Analysis of ABCG2 expression and side population identifies intrinsic drug efflux in the HCC cell line MHCC-97L and its modulation by Akt signaling

Transcription

1 Carcinogenesis vol.29 no.12 pp , 2008 doi: /carcin/bgn223 Advance Access publication September 26, 2008 Analysis of ABCG2 expression and side population identifies intrinsic drug efflux in the HCC cell line MHCC-97L and its modulation by Akt signaling Chen Hu y, Hong Li 1,y, Jinjun Li 1,, Zheng Zhu 1, Shengyong Yin, Xiangfang Hao 1, Ming Yao 1, Shusen Zheng and Jianren Gu 1 Department of General Surgery, the First Affiliated Hospital, School of Medicine, Zhejiang University, 79 Qingchun Road, Hangzhou , People s Republic of China and 1 State Key Laboratory of Oncogenes and Related Genes, Shanghai Cancer Institute and Cancer Institute of Shanghai Jiao Tong University, 25/Ln 2200 Xietu Road, Shanghai , People s Republic of China To whom correspondence should be addressed. Tel: þ ; Fax: þ ; jjli@shsci.org Correspondence may also be addressed to Shusen Zheng. Tel: þ ; Fax: þ ; shusenzheng@zju.edu.cn Active drug efflux by the adenosine triphosphate-binding cassette (ABC) transporter ABCG2 is one of the common mechanisms causing multiple drug resistance in various human cancers. In the intrinsic drug resistance of hepatocellular carcinoma (HCC), the role of ABCG2 is closely associated with side population (SP), a minor subset of cancer stem-like cells with unique capacity to extrude lipophilic dye Hoechst and many chemotherapeutic agents. In this study, we showed that ABCG2 was intrinsically expressed in a subgroup of HCC tissues and its expression pattern significantly influenced the levels of drug efflux from HCC cell lines. In MHCC-97L HCC cell line with intrinsic ABCG2 expression, we confirmed the importance of SP cells to the drug effluxrelated chemotherapy resistance and found that the SP analysis provided an efficient method to evaluate the functional activity of ABCG2 transporter. In this cell line, we discovered that the SP proportion was modulated by the treatments of Akt signaling inhibitors and serum supplement, which led to the finding that Akt signaling was able to regulate the SP cells efflux activity via altering the subcellular localization of ABCG2 transporter. We further demonstrated that the Akt signaling inhibition attenuated the doxorubicin efflux from MHCC-97L cells and increased the drug efficacy. Our results indicate the protective role of intrinsic ABCG2 expression in HCC cells and suggest that suppressing Akt signaling could help overcome the drug efflux by ABCG2 transporter. Introduction Hepatocellular carcinoma (HCC) is one of most common cancers worldwide, and its resistance to conventional drugs is a major impediment to successful chemotherapy (1,2). Chemotherapy resistance can be developed by various mechanisms, including altered drug target, enhanced detoxication capacity, impaired cell death pathway and reduced drug accumulation in cancer cells (3). In HCC, a leading cause for chemotherapeutic failure has been attributed to the fact that a large proportion of cancer cells express relatively high levels of several adenosine triphosphate-binding cassette (ABC) transporters (2,4), which actively pump out a broad spectrum of clinically relevant compounds and decrease the intracellular drug accumulation (5). The ABCG2 transporter, first identified in a doxorubicin-selected MCF-7 breast cancer Abbreviations: ABC, adenosine triphosphate-binding cassette; DMEM, Dulbecco s modified Eagle s medium; FBS, fetal bovine serum; FTC, fumitremorgin C; HCC, hepatocellular carcinoma; mtor, mammalian target of rapamycin; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; PI3K, phosphoinositide 3-kinase; sirna, small-interfering RNA; SP, side population. y These authors contributed equally to this work. cell line, has been shown to mediate multiple drug resistance in various cancers (6,7); its substrates include a variety of anticancer agents such as mitoxantrone, camptothecins, anthracyclines, flavopiridol and antifolates (7). In HCC, ABCG2 expression is known to be preferentially upregulated in the stem/stem-like cell compartment (8,9), suggesting a role of ABCG2 in protecting this population against chemotherapy. Recently, the role of ABCG2 was underlined with the identification of side population (SP) cells in many cancers (10,11). The SP, a rare subset of cells distinguished by their low Hoechst dye staining in flow cytometry, was first described by Goodell et al. (12) in identifying hematopoietic stem cells in bone marrow. Unlike the common methods that involve recognition of cell surface markers, SP phenotype is functionally defined by the cells performance in pumping out the DNA-binding dye Hoechst (11). Studies in leukemia and some solid tumors have shown that the SP phenotype could be used to enrich cancer stem cells (10,13 15), although it appeared heterogenous in terms of cell surface marker profile (11). Experimental evidence from the Abcg2 / knockout mice model has demonstrated that ABCG2 is the primary transporter mediating the SP phenotype, and several other ABC family members have overlapping function in Hoechst dye efflux (16,17). The SP cells have been shown to possess extreme tumorigenicity and intrinsic drug resistance, suggesting that the existence of this population could be an important cause of the tumor recurrence after chemotherapy. Currently, the mechanism leading to the upregulated ABCG2 expression in SP cells remains unrevealed, and the direct relationship between ABCG2 expression and the stem cell properties is also undetermined (7,18). In HCC cell lines, previous studies have reported the existence of a distinct SP with cancer stem/stem-like cell properties (19,20). However, ABCG2 expression, SP cells and their relevance to chemotherapy resistance in HCC still need further investigation. HCC is known as a cancer with frequent occurrence of intrinsic drug resistance, and one common cause is that the bulk tumor cells express ABC pumps, resulting in that the tumor shows little response to conventional chemotherapy (2,21). It is necessary to explore the existence of this intrinsic ABCG2 expression pattern in HCC primary tumors, although ABCG2 expression is known to be mainly present in a rare stem/ progenitor compartment (9). In addition to the expression level, the functional status of ABC transporters is also an important factor in the drug efflux process. SP analysis has often been used to identify stem/ stem-like cells, but it also provides a good method to evaluate the activity of ABCG2 transporter (18). In murine bone marrow, it was found that the ABCG2 activity of SP cells could be modulated by Akt signaling (22). Whether there is a similar functional regulation in HCC SP cells remains an intriguing question. In the present study, we investigated the ABCG2 expression patterns in HCCs and demonstrated that the ABCG2 expression conferred the efflux capacity to HCC cells, especially to the SP cells. In the MHCC-97L cell line with intrinsic ABCG2 expression, we found that the SP cells not merely possessed greater drug efflux capacity but also exhibited higher Akt signaling activity than the non-sp cells. Further studies showed that Akt signaling could significantly modulate the SP proportion in the MHCC-97L cell line, and the regulatory effect could be achieved by promoting the subcellular translocation of ABCG2 transporter. For therapeutic implication, we demonstrated that the inhibition of Akt signaling in MHCC-97L cells increased intracellular doxorubicin retention, which could potentiate the chemotherapeutic effect on HCC cells. Materials and methods HCC samples and tissue array Fifty-five human HCC tissue specimens were collected from patients who underwent surgical treatment at Qidong Liver Cancer Institute (Qidong, Ó The Author Published by Oxford University Press. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org 2289

2 C.Hu et al. China) or at the First Affiliated Hospital of Zhejiang University (Hangzhou, China). The 55 HCC patients comprised 47 males and 8 females (mean age: 51.7 years, ranging from 31 to 74 years), and no patient received preoperative chemotherapy. All the procedures were performed under consensus agreements and in accordance with the China Ethical Review Committee. Collected samples were fixed in 4% phosphate-buffered formalin (ph 7.4) for at least 72 h and routinely embedded in paraffin. Tissue array was constructed as described previously (23). Cell culture and reagents The HCC cell lines Huh7 and Hep3B were provided by Cell Bank of Chinese Academy of Sciences (Shanghai, China), and the MHCC-97L cell line was from Zhongshan Hospital, Fudan University (Shanghai). These cell lines were cultured in Dulbecco s modified Eagle s medium (DMEM) (Sigma, St Louis, MO) containing 10% heat-inactivated fetal bovine serum (FBS) (Hyclone, Logan, UT) and incubated at 37 C under a humidified atmosphere with 5% CO 2. Antibodies used in this study are listed in supplementary Table I (available at Carcinogenesis Online). Phosphoinositide 3-kinase (PI3K) inhibitor LY and mammalian target of rapamycin (mtor) inhibitor rapamycin were from Cell Signaling Technology (Danvers, MA). Chemicals and other reagents without specific mention were all purchased from Sigma. Immunostaining of HCC tissue array and cultured cells Paraffin-embedded HCC sections (5 lm in thickness) were dewaxed in xylene twice for 10 min each and were rehydrated using a graded series of alcohol. Endogenous peroxidase was inactivated by incubating sections in methanol (with 0.3% hydrogen peroxide) for 30 min at room temperature. Antigen retrieval was performed by heating slides in sodium-citrate buffer (10 mm, ph 6.0) at 94 C for 10 min. Cultured HCC cells were maintained in slide chambers for at least 48 h and fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) (ph 7.4) for 20 min at room temperature. For analysis of subcellular ABCG2 localization, fixed cells were permeabilized in PBS with 0.02% Triton X-100 prior to primary antibody incubation. Immunodetection was performed as described previously (24) (optimal dilutions for primary antibodies refer to supplementary Table I, available at Carcinogenesis Online). The results were observed and photographed under an Axioskop 2 microscope (Carl Zeiss, Oberkochen, Germany) with a DP70 CCD system (Olympus, Tokyo, Japan). Western blotting Cell lysates were prepared in radioimmunoprecipitation assay buffer (Santa Cruz Biotechnology, Santa Cruz, CA) with 1 mm sodium fluoride, 2 mm sodium orthovanadate and a cocktail of proteinase inhibitors (Santa Cruz Biotechnology). Total protein concentration was determined by the modified Bradford assay according to the manufacturer s instruction (Sigma). Quantified cell lysates (30 60 lg protein loaded for each lane) were resolved by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA). The electroblotted membranes were blocked with PBS containing 5% non-fat milk (Santa Cruz Biotechnology) and then probed with the primary antibodies overnight at 4 C (refer to supplementary Table I, available at Carcinogenesis Online). After washing, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (Sigma) at room temperature for 1 h, followed by chemiluminescence detection using SuperSignal West Femto Maximum Sensitivity Substrate Kit (Pierce, Rockford, IL). Finally, the membranes were exposed to X-ray film (Kodak, Rochester, NY) for the desired time. Real-time reverse transcription polymerase chain reaction Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA) and reversely transcribed using oligo (dt) primer and Moloney murine leukemia virus reverse transcriptase (TaKaRa Biotechnology, Shiga, Japan). Real-time polymerase chain reaction (PCR) was subsequently performed using the ABI Prism 7300 Sequence Detection System (Applied Biosystems, Foster City, CA) at the recommended thermal cycling settings: one initial cycle at 95 C for 10 s followed by 40 cycles of 5 s at 95 C and 31 s at 60 C. Each 20 ll PCR contained 2 lldiluted complementarydna(50ngstartingtotalrna)and0.2lm corresponding primers in mixed SYBR Premix Ex Taq Solutions (TaKaRa Biotechnology). The primers for target genes are listed as follows: ABCB1 (forward: 5#-GCTCCTGACTATGCCAAAGC-3# and reverse: 5#-TCTTC- ACCTCCAGGCTCAGT-3#), ABCC1 (forward: 5#-CTGGGCTTATTTCGGA- TCAA-3# and reverse: 5#-TGAATGGGTCCAGGTTCATT-3#), ABCG2 (forward: 5#-CACCTTATTGGCCTCAGGAA-3# and reverse: 5#-CCTGCTT- GGAAGGCTCTATG-3#) and ABCA3 (forward: 5#-AGAAATACGGTGCC- GGCTATCACA-3# and reverse: 5#-CAATGCCCAGCTCTTTCTGCTTCT-3#). The expression levels were normalized to the internal reference gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) SP analysis The SP analysis procedures were based on the protocol established by Goodell et al. (12). Briefly, cells were trypsinized from dishes and were suspended at cells per ml in DMEM with 2% FBS and 10 mm N-2-hydroxyethylpiperazine-N#-2-ethanesulfonic acid. These cells were stained at 37 C for 90 min with 5 lg/ml Hoechst (Invitrogen), either alone or in the presence of 10 lm fumitremorgin C (FTC) as a control. The incubation was performed in water bath with shaking at intervals. At the end of incubation, the cells were centrifuged at 4 C and resuspended in ice-cold PBS containing 2 lg/ml propidium iodide. The flow cytometry analysis and the fluorescence-activated cell sorting were performed using Epics Altra flow cell sorter (Beckman Coulter, Fullerton, CA) equipped with a 488 nm argon laser and an INNOVA 90-CA5 UV laser (Coherent, Santa Clara, CA). The Hoechst dye was excited by a 351 nm ultraviolet laser and the fluorescence emission was collected through a 450DF20 filter (Hoechst blue) and a 675ALP filter (Hoechst red). In the present study, the SP gate in flow cytometry analysis was defined as the diminished area on the dot plot in the presence of FTC. To examine the clonogenicity of the cancer cells with drug treatment, the SP and non-sp MHCC-97L cells were sorted in the presence of 500 ng/ml doxorubicin and seeded into 96-well plates in replicates at 20 cells per well. The cells were cultured in DMEM with 10% FBS, and the number of colonies was counted after 14 days. Doxorubicin efflux analysis by flow cytometry To analyze doxorubicin efflux from the cells, intracellular doxorubicin fluorescence was measured by flow cytometry as described previously (25). Briefly, cultured cells were first exposed to 10 lg/ml doxorubicin in DMEM at 37 C for 60 min. After exposure and washing, the cells were released in drug-free 10% FBS DMEM for 90 min and then subjected to flow cytometry. When assessing ABCG2-mediated doxorubicin efflux, FTC was added in the incubation medium at a concentration of 10 lm for ABCG2-inhibited control. Intracellular doxorubicin was excited by a 488 nm argon laser and the fluorescence was measured through a 575BP filter. To measure the efflux activity of SP cell, the samples were coincubated with doxorubicin (500 ng/ml) and Hoechst for 60 min when in the staining process and then were transferred to drug-free staining buffer for another 30 min incubation with Hoechst dye alone (10,13). RNA interference-based gene knockdown experiment Small-interfering RNA (sirna) oligo for Akt1 and general negative control (Cat. no. B01001) was synthesized and annealed by GenePharma (Shanghai, China). Three fragments were designed to target Akt1 transcripts, and the silencing effect of the following sequence was validated through western blotting: 5#-CGAGGGGAGUACAUCAAGAtt-3# (sense) and 5#-UCUUGAU- GUACUCCCCUCGtt-3# (antisense). RNA interference was performed using the reverse transfection method: prior to cell plating, sirna and Lipofectamine 2000 (Invitrogen) in Opti-MEM (Invitrogen) were mixed and incubated in 12-well plates (40 pmol RNA interference molecules and 2 ll Lipofectamine 2000); cells were trypsinized into suspension at cells per ml and plated in medium containing the prepared RNA oligo-lipofectamine 2000 complex. The cells were cultured at 37 C for 48 h and then harvested for further analysis. Statistical analysis Statistical analyses were performedusing SPSS 13.0 software (SPSS, Chicago, IL). Continuous data were presented as the mean ± SD and compared using the unpaired Student s t-test. The v 2 or Fisher s exact test was used for categorical data where appropriate. P, 0.05 was considered a statistically significant difference. Additional information can be found in supplementary Materials and methods (available at Carcinogenesis Online). Results Intrinsic ABCG2 expression exists in a subset of primary HCC tissues To characterize the ABCG2 expression patterns, the immunohistological analysis was performed in a tissue array, which included 55 tumors from HCC patients without chemotherapy prior to surgical treatment. The specificity of the immunodetection was confirmed using two monoclonal antibodies (5D3 and BXP-21). Of the 55 HCC samples, ABCG2 expression was detected in 42 cases and absent from 13 cases (Figure 1A). In the ABCG2 þ HCC tissues, we found that its staining patterns could be classified into different types. In agreement with previous reports (9), ABCG2 expression was present in a minor cell population (,25% of the cancer cells) of 29 cases (52.7% of the HCC cases), and its distribution was scattered or focally clustered (Figure 1Aa and b). However, we found a distinct diffuse pattern in another 13 HCC cases (23.6% of the HCC cases), in which a large proportion of ABCG2 þ cells (.50% of the cancer cells) were

3 Side population and intrinsic ABCG2 expression in HCC Fig. 1. ABCG2 expression and SP in human HCC tissues and cell lines. (A) Representative ABCG2 immunostaining in a HCC tissue array containing 55 samples: ABCG2-positive cells are scattered or focally clustered in 29 cases (a and b); diffuse staining pattern is present in 13 cases (c and d) and no positive immunostaining is detected in 13 cases (e and f). The number of cases in each subset is shown in the left corner of a, c and e. Original magnifications: (a, c and e) 50 and (b, d and f) 400. (B) Immunofluorescence analysis of ABCG2 in MHCC-97L, Huh7 and Hep3B cell lines. ABCG2 was probed using a monoclonal primary antibody (5D3) and labeled by Alexa594-conjugated secondary antibody (red). Nuclei were counterstained by 4#,6-diamidino-2-phenylindole (blue). (C) Relative messenger RNA (mrna) expression of ABCG2 was determined by real-time reverse transcription PCR in MHCC-97L, Huh7 and Hep3B cell lines. Expression levels were normalized to GAPDH. (D) Western blotting of ABCG2 in MHCC-97L, Huh7 and Hep3B HCC cell lines. b-actin was examined as a loading control. (E) Representative SP analysis of MHCC-97L, Huh7 and Hep3B cell lines (n 5 3) (described in Materials and Methods). The flow cytometry gate for SP cells is defined by the diminished area (the framed area) in the presence of 10 lm FTC. (F) Flow cytometry histograms of doxorubicin efflux assay for MHCC-97L, Huh7 and Hep3B HCC cell lines (n 5 3). ABCG2-mediated doxorubicin efflux was represented by the fluorescence difference between the FTCtreated cells and the -untreated cells. Gray dashed line, control cells without doxorubicin incubation; red line, cells incubated with 10 lg/ml doxorubicin alone and blue line, cells incubated with doxorubicin in the presence of FTC (10 lm). (G) Cytotoxicity assay for HCC cell lines with doxorubicin treatment. Doxorubicin concentration ranged from to 50.0 lg/ml. Relative cell viability was determined by the 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (MTT) assay. (H) Cytotoxic effect of doxorubicin (1.0 lg/ml) on HCC cell lines in the presence or absence of 10 lm FTC (n 5 3). Relative cell viability was determined using the MTT assay. P, 0.05 t-test. intensely stained and spread over the parenchyma (Figure 1Ac and d). The diffuse expression pattern demonstrated that ABCG2 could be intrinsically overexpressed in the HCC tissues. With respect to clinicopathologic features, we found that the diffuse intrinsic ABCG2 expression was preferentially present in HCC with liver cirrhosis (Table I) (P, 0.001), indicating that this expression pattern is correlated with the etiological background of HCC. SP analysis and ABCG2 functional assay in HCC cell lines with different ABCG2 expression patterns In human HCC cell lines, we found that MHCC-97L cell line displayed diffuse ABCG2 immunostaining as that observed in the HCC tissue array, whereas in Huh7 and Hep3B cell lines ABCG2 expression was rarely detected (Figure 1B). To verify the ABCG2 expression levels of the HCC cell lines, we performed real-time reverse transcription PCR and western blotting experiments and obtained consistent results (Figure 1C and D). To assess the impact of ABCG2 expression patterns on the SP proportion, we analyzed MHCC-97L, Huh7 and Hep3B HCC cell lines. In the three cell lines, we all detected the SP fraction, which was characterized by a low fluorescent tail in the Hoechst blue versus red scatter plot. As shown in Figure 1E, MHCC-97L cell line possessed much larger SP proportion (5 10%) than Huh7 and Hep3B cell lines (both,1%), indicating that the high-level ABCG2 expression could significantly raise the SP fraction. To confirm the major contribution of ABCG2 in the SP phenotype, we compared the effects of transporter-specific inhibitors on blocking the Hoechst efflux from MHCC-97L cells. The results showed that the MHCC-97L SP 2291

4 C.Hu et al. Table I. Clinicopathologic characteristics of HCC specimens with different ABCG2 immunostaining pattern Variables ABCG2 staining pattern P value Diffuse n (%) (n 5 13) Scattered n (%) (n 5 29) Negative n (%) (n 5 13) Tumor size (cm),5 9 (69.20) 12 (41.40) 4 (30.80) (30.80) 17 (58.60) 9 (69.20) TNM staging I II 11 (84.60) 20 (69.00) 9 (69.20) III IV 2 (15.40) 9 (31.00) 4 (30.80) Edmondson grading I II 7 (53.80) 11 (37.90) 3 (23.10) III IV 6 (46.20) 18 (62.10) 10 (76.90) Cirrhosis Absent 1 (7.70) 20 (69.00) 12 (92.30) 2.2E-05 a Present 12 (92.30) 9 (31.00) 1 (7.70) Serum AFP (ng/ml) 20 4 (30.80) 6 (20.70) 5 (38.50) (69.20) 23 (79.30) 8 (61.50) Serum HBsAg Absent 1 (7.70) 8 (27.60) 4 (30.80) Present 12 (92.30) 21 (72.40) 9 (69.20) P value represents the probability from v 2 test for ABCG2 staining pattern between variable subgroups. TNM, TNM classification of malignant tumours; AFP, alpha-fetoprotein; HBsAg, hepatitis B virus surface antigen. a v 2 test, P, fraction could be completely abrogated by the ABCG2-specific inhibitor FTC but was resistant to verapamil (ABCB1 inhibitor) and probenecid (ABCC1 inhibitor) (supplementary Figure S1 is available at Carcinogenesis Online). We also evaluated the role of ABCG2 in mediating the drug efflux of the HCC cell lines. The anticancer drug doxorubicin, which is commonly used in HCC chemotherapy, is a naturally fluorescent molecule that can be transported by ABCG2. The intracellular concentration of doxorubicin can be monitored by flow cytometry as described previously (26). By the use of the ABCG2-specific inhibitor FTC, the ABCG2-mediated doxorubicin efflux could be represented by the fluorescence intensity difference between FTC-treated cells and -untreated cells. As shown in Figure 1F, the FTC-inhibitable doxorubicin efflux was observed in the MHCC-97L cells but not detected in Huh7 and Hep3B cell lines. Moreover, we examined the cytotoxic effect of doxorubicin on the HCC cell lines. The MHCC-97L cell line was resistant to a higher level of doxorubicin compared with the Huh7 and Hep3B cell lines (Figure 1G), and FTC was able to sensitize MHCC-97L cells to doxorubicin treatment (Figure 1H). These results clearly indicated that the active drug efflux by intrinsic ABCG2 expression greatly contributed to the drug resistance of MHCC-97L cells. Characterization of SP and non-sp cells in MHCC-97L cell line with intrinsic ABCG2 expression Using SP analysis and fluorescence-activated cell sorting techniques, we isolated SP and non-sp cells from MHCC-97L cell line (Figure 2A) and examined their growth patterns. In contrast to previous studies that reported HCC SP cells with extreme tumorigenicity (19,20), the cell cycle analysis and clonogenicity assay showed no significant difference between the SP and non-sp MHCC-97L cells under normal culture conditions (Figure 2B and C). To confirm that MHCC-97L SP cells can extrude drug more actively than non-sp cells, we measured the doxorubicin efflux from the two populations simultaneously when performing SP analysis (13). We observed that the doxorubicin retention in the SP cells was much less than that in the non-sp cells (Figure 2D), indicating that the drug was pumped out more efficiently in the SP cells. We further isolated SP and non-sp MHCC-97L cells in the presence of doxorubicin and 2292 compared their clonogenicity. The MHCC-97L SP cells formed more colonies than non-sp cells after the exposure to doxorubicin, suggesting that the more powerful efflux activity in SP cells could better protect them against the cytotoxic insult (Figure 2E). To determine whether the SP phenotype-related ABC family members were differentially expressed between the MHCC-97L SP and non- SP fraction, real-time reverse transcription PCR was conducted to compare their expression levels. As shown in Figure 2F, we confirmed that the MHCC-97L SP cells had more ABCG2 expression than the non-sp cells (SP:non-SP 5 2-fold), but it was noteworthy that non-sp cells also expressed considerable amount of ABCG2, indicating that the ABCG2 expression in the SP/non-SP cells was not absolutely turned on or off. Besides ABCG2, the expression levels of ABCB1 and ABCC1 in the SP cells were also higher than those in the non-sp cells, indicating that multiple ABC transporters were enriched in the SP cells. However, the expression of intracellular membrane transporter ABCA3 was detected at a very low level in both the populations (Figure 2F). Given the ABCG2 expression pattern in the SP and non-sp MHCC- 97L cells, we wondered whether there were other mechanisms involved in regulating the SP phenotype. Prompted by the previous study in hematopoietic cells (22), we assessed the activity of the PI3K/Akt/ mtor pathway and the mitogen-activated protein kinase pathway in the SP and non-sp MHCC-97L cells. Interestingly, the results revealed that SP cells exhibited higher phosphorylation levels of Akt and S6 protein compared with non-sp cells, indicating that Akt signaling could contribute to the SP phenotype (Figure 2G). However, no significant difference was observed in the levels of phosphorylated extracellular signal-regulated kinase. Considering that the low serum concentration in the routine Hoechst incubation (2% FBS) could reduce the overall cell signaling activity, we also performed western blot analysis of SP/ non-sp cells incubated in 10% FBS Hoechst staining medium, and the result was consistent with that in the 2% FBS condition (Figure 2G). Inhibition of Akt signaling attenuates SP fraction in MHCC-97L cell line To explore the possible contribution of Akt signaling in the SP phenotype, we examined the SP proportion of MHCC-97L cells when the PI3K inhibitor LY or the mtor inhibitor rapamycin was included in Hoechst incubation buffer. The SP fraction in MHCC-97L cells significantly decreased by 93.2% with LY (5 lm) and by 70.5% with rapamycin (5 lm), suggesting that blocking Akt signaling could efficiently attenuate the efflux activity of MHCC-97L SP cells (Figure 3A and B). We monitored the apoptosis rate of MHCC-97L cells simultaneously when performing the SP analysis and confirmed that the reduced SP fraction was not caused by the non-specific toxicity of the treatment. With regard to the mitogen-activated protein kinase signaling, the treatment with the mitogen-activated protein kinase inhibitor PD98059 (5 lm) had no significant effect on the SP fraction of the MHCC-97L cell line (supplementary Figure S2 is available at Carcinogenesis Online). As Akt signaling is known to be susceptible to serum stimulation, we increased the serum concentration (10% FBS) in the Hoechst incubation medium to observe the effect on SP proportion. Stimulated by the serum supplement, MHCC-97L SP fraction increased 1.95-folds, and the inhibitory effect of LY and rapamycin was alleviated (Figure 3A and B). Figure 3C presents the western blotting result verifying the changes in Akt signaling activity under the treatments of inhibitors and serum stimulation. It showed that the activity of Akt signaling was closely associated with the SP proportion in MHCC-97 cells. Although we had assumed that the modulation of SP proportion would depend on regulating ABCG2 expression, no evident downregulation of ABCG2 expression was observed in MHCC-97L cells treated with LY or rapamycin (Figure 3C). We also used sirna method to verify the effect of knockdowned Akt1 expression on the SP fraction. As shown in Figure 3D, the SP proportion in the Akt1-knockdowned MHCC-97L cells markedly decreased by 71.7% compared with the negative sirna control. However, the ABCG2 expression level was nearly unchanged by the RNA interference (Figure 3E).

5 Side population and intrinsic ABCG2 expression in HCC Fig. 2. Properties of SP and non-sp from MHCC-97L cell line with intrinsic ABCG2 expression. (A) Fluorescence-activated cell sorting isolation of SP and non- SP cells from the MHCC-97L cell line. The purity of fluorescence-activated cell sorting isolation (Pre-Sort) was confirmed by immediate reanalysis of sorted SP cells and non-sp cells. (B) Cell cycle analysis of sorted SP (left) and non-sp cells (right) from MHCC-97L cell line at day 3 after fluorescence-activated cell sorting isolation (n 5 3). No significant difference was observed in cell cycle distribution between SP and non-sp cells. (C) Colonies formed by sorted SP and non-sp MHCC-97L cells after 2 week culture (n 5 3): colonies in plate wells after Giemsa staining (left); representative colonies on plastic plates (middle) and in 3D Matrigel (right). (D) Flow cytometry histograms of doxorubicin efflux from MHCC-97L SP and non-sp cells. Doxorubicin fluorescence was plotted for the gated SP cells (red line) and non-sp cells (blue line). (E) Clonogenicity assay for SP and non-sp MHCC-97L cells with exposure to doxorubicin (500 ng/ml) during SP analysis (n 5 3). Sorted SP and non-sp cells were seeded in 96-well plates at a density of 20 cells per well, and the number of colonies was counted after 2 weeks. SP cells showed survival advantage over non-sp cells. P, 0.05, t-test. (F) Relative messenger RNA expression of ABCB1, ABCC1, ABCG2 and ABCA3 in SP and non-sp MHCC-97L cells. Expression levels were normalized to GAPDH. P, 0.05, t-test. (G) Western blotting of phosphorylated/total Akt, phosphorylated/total ribosomal protein S6 and phosphorylated/total extracellular signal-regulated kinase (Erk1/2) in sorted SP and non-sp MHCC-97L cells. NS, non-sp. The experiments were done for both the 2% FBS and 10% FBS incubation conditions. 2293

6 C.Hu et al. Fig. 3. Inhibition of Akt signaling attenuates Hoechst efflux of MHCC-97L SP. (A) Flow cytometry plots of SP analysis for MHCC-97L cells with treatments of Akt signaling inhibitors and serum supplement (n 5 3). In staining buffer containing 2% FBS or 10% FBS, MHCC-97L cells were incubated with Hoechst in the presence of LY (5 lm) or rapamycin (5 lm) and then subjected to SP analysis. Dimethyl sulfoxide (DMSO) was added as control. (B) Plots for SP percentage of MHCC-97L cells with indicated treatments. P, 0.05, t-test. (C) Western blotting of Akt signaling mediators and ABCG2 expression in MHCC-97L cells treated with LY (5 lm) or rapamycin (5 lm) in staining buffer containing 10% FBS or 2% FBS. (D) SP analysis of MHCC- 97L cells in Akt1 sirna experiments (n 5 3). Gene knockdown was performed using Akt1 sirna transfection, and a general negative sirna oligo was used as negative control. (E) Western blot analysis for Akt1 and ABCG2 expression of MHCC-97L cells in Akt1 sirna experiments. The silencing effect of Akt1 sirna was validated by decreased levels of Akt1 protein. b-actin was examined as a loading control. Akt signaling inhibition alters subcellular distribution of ABCG2 in MHCC-97L SP cells To investigate the effect of Akt signaling on ABCG2 transporter, we also analyzed the ABCG2 expression in isolated SP and non-sp MHCC-97L cells separately. Like the experiment performed in the whole MHCC-97L cell population, either the LY or rapamycin treatment had minimal effect on ABCG2 expression in both the SP and non-sp cells (Figure 4A), suggesting that the modulation of efflux activity in MHCC-97L cells did not require changing the amount of overall cellular ABCG2 protein As the Akt signaling in hematopoietic cells can regulate SP phenotype via ABCG2 subcellular translocation (22), we investigated whether this mechanism exists in the HCC cells. The freshly isolated MHCC-97L SP cells were cultured with or without Akt signaling inhibitors for 90 min, and then the ABCG2 subcellular localization was examined by immunostaining. Interestingly, the LY or rapamycin treatment did reduce the ABCG2 staining on the plasma membrane, suggesting that the downregulated Akt signaling could be responsible for the loss of functional ABCG2 on the plasma membrane (Figure 4B).

7 Side population and intrinsic ABCG2 expression in HCC LY and rapamycin increase intracellular doxorubicin retention in MHCC-97L cells and enhance drug efficacy To test whether Akt signaling inhibitors could also attenuate drug efflux of ABCG2 transporter, we evaluated the effect of LY and rapamycin on the doxorubicin efflux from MHCC-97L cells. As measured by flow cytometry, intracellular doxorubicin in the cells treated with LY or rapamycin persisted at a high level after release in the drug-free medium; in contrast, the intracellular doxorubicin levels rapidly decreased in the untreated control cells (Figure 5A). We further assessed the efficacy of cytotoxic drugs in combination with Akt signaling inhibitors. The MTT assay demonstrated that LY or rapamycin was able to sensitize the MHCC-97L cells Fig. 4. Effect of Akt signaling on the expression and subcellular localization of ABCG2 in MHCC-97L SP cells. (A) Western blotting of ABCG2 expression and Akt signaling in fluorescence-activated cell sorting-sorted SP and non-sp MHCC-97L cells. Freshly isolated SP and non-sp cells were maintained in 10% FBS DMEM for 90 min at 37 C, with or without Akt signaling inhibitor (5 lm LY or 5 lm rapamycin) and then subjected to western blot analysis. ABCG2 expression level in SP or non-sp cells is nearly unchanged with Akt signaling inhibition. (B) Immunofluorescence analysis of subcellular ABCG2 expression in SP and non-sp MHCC-97L cells. Before immunostaining analysis, freshly isolated SP cells were incubated in 10% FBS DMEM for 90 min at 37 C, in the presence or absence of Akt signaling inhibitor (5 lm LY or 5 lm rapamycin). BXP-21 monoclonal ABCG2 antibody was used for the immunodetection, and the nuclei were stained by 4#,6-diamidino-2-phenylindole (DAPI). Fig. 5. LY (LY) and rapamycin (Rapa) reduce doxorubicin (Doxo) efflux in MHCC-97L cells and enhance drug efficacy. (A) Flow cytometry histogram of intracellular doxorubicin fluorescence in MHCC-97L cells treated with LY or rapamycin. Gray dotted line, cells without doxorubicin treatment; green line, cells with doxorubicin only; blue line, cells with doxorubicin and 5 lm LY294002; red line, cells with doxorubicin and 5 lm rapamycin and black line: cells with doxorubicin and 10 lm FTC. (B) Relative viability of MHCC-97L cells with doxorubicin treatment alone or in combination with Akt signaling inhibitors (n 5 4). Doxorubicin concentration ranged from 0.5 to 8.0 lg/ml. The 50% inhibitory concentration for each doxorubicin treatment was 4.0 lg/ml (doxorubicin alone), 2.5 lg/ml (doxorubicin þ 5 lm rapamycin) and 1.3 lg/ml (doxorubicin þ 5 lm LY294002). (C) Flow cytometry plots of Annexin Vand 7-amino-actinomycin D (7-AAD) apoptosis assay for MHCC-97L cells in doxorubicin treatment (percentage of dead and apoptotic cells was shown in the plots). a, control (DMSO); b, 5 lm rapamycin; c, 5 lm LY294002; d, 2 lg/ml doxorubicin; e, 2 lg/ml doxorubicin þ 5 lm rapamycin and f, 2 lg/ml doxorubicin þ 5 lm LY (D) Hypothetical diagram for SP cells in the MHCC-97L cell line. In this cell line, a large proportion of cells intrinsically express ABCG2, and the SP fraction was enriched with efflux capacity (dashed line, threshold for determining SP). In this context, cancer stem cells might be overwhelmed by the ABCG2 þ Hoechst low cells, and the Hoechst staining might fail to define an accurate stem-like population. Moreover, there can be transitions between the SP and non-sp phenotypes when the efflux capacity of cancer cells is modulated. In addition to ABCG2 expression level, other mechanisms, including cellular signaling activity, could also influence the cells efflux performance of SP cells. 2295

8 C.Hu et al. to doxorubicin treatment (Figure 5B). Using the Annexin V 7-AAD double-staining assay in MHCC-97L cells, we also observed that the apoptosis rate induced by doxorubicin (2 lg/ml) was increased in the presence of the Akt signaling inhibitors (Figure 5C). Discussion In clinical cancer treatment, the intrinsic expression of ABC transporters is often associated with tumor s inherent drug resistance, in which most of the cancer cells are naturally resistant to anticancer agents even without an initial response at the start of chemotherapy (3,21). In primary HCC tissues, we showed that ABCG2 expression patterns manifested different types. The rare expression pattern was in agreement with that described by the cancer stem cell model (9), in which ABCG2 was preferentially expressed by a rare subset of stem/ stem-like SP cells, facilitating their resistance to conventional chemotherapy (21). However, our findings highlighted the diffuse expression pattern, and this pattern could be closely associated with the situation in which the bulk of HCC cells are resistant to chemotherapy. In addition, we found that this expression pattern correlated with the occurrence of liver cirrhosis, suggesting that aberrant ABCG2 expression would arise in certain pathological states (9). In the present study, the MHCC-97L cell line, which showed high levels of ABCG2 expression, provided an experimental model to study the role of ABCG2 and SP fraction in the HCC cells. We confirmed that the MHCC-97L SP cells possessed more ABCG2 expression and greater efflux activity, indicating that this population could significantly contribute to the drug resistance of HCC cells. Taking advantage of the sensitivity of SP analysis, we also detected rare SP cells in Hep3B and Huh7 cell lines, although the contribution of this minor population in drug efflux was masked by the rest of HCC cells. Moreover, we found that the SP analysis could reflect the actual efflux capacity of cancer cells with ABCG2 expression. Although the MHCC-97L SP cells showed survival advantage over non-sp cells when exposed to drug treatment, the two populations exhibited similar growth patterns in normal culture conditions. This result appeared inconsistent with the previous study that reported HCC SP cells possessed cancer stem cell properties (19), but it was in agreement with the recent observation in some other cancer cell lines (27 29). One possible explanation for this discrepancy could be that the relevance of the SP phenotype and stemness could vary among cancer cells with different ABCG2 expression patterns. The intrinsic ABCG2 expression in MHCC-97L cell line increases efflux activity in the bulk cancer cells, resulting in that the SP analysis is insufficient to define an accurate cancer stem cell population. As illustrated in Figure 5D, the rare stem-like cells in the SP fraction are greatly outnumbered by the bulk ABCG2 þ Hoechst low cells. PI3K/Akt/mTOR signaling is composed of key mediators that can trigger a cascade of cellular responses, including cell growth, metabolism, mobility and survival (30). It is well known that the Akt signaling confers drug resistance by suppressing apoptotic pathways, but its role in regulating drug efflux from cancer cells is not well understood. We found that the Akt signaling in MHCC-97L SP cells could greatly modulate their efflux capacity, as reflected by the changes of SP proportion under the treatments of Akt signaling inhibitors and serum stimulus. This result suggested, in the context of intrinsic ABCG2 expression, that SP and non-sp fraction are not mutually exclusive since there can be transitions between the two phenotypes when the efflux capacity is regulated by cellular signaling (Figure 5D). It also indicates that the intracellular signaling and external stimulus (for example, from serum and tumor microenvironment) could affect the drug efflux from cancer cells and ultimately influence the chemotherapy outcome. One example in our study is that LY and rapamycin could attenuate the doxorubicin efflux from MHCC-97L cells by suppressing Akt signaling, highlighting the importance of intracellular signaling in the drug efflux process. We demonstrated that the modulation of SP phenotype was associated with the functional regulation of ABCG2 transporter, which was achieved by the alteration of its subcellular distribution. As suggested 2296 by previous studies in other ABC family members (31), the regulation of intracellular ABC transporter trafficking involves complex processes including cellular energy metabolism, cytoskeletal arrangement and vesicular targeting. The detailed manner needs further investigations. Although we demonstrated the subcellular ABCG2 translocation under Akt signaling inhibition, it did not exclude other mechanisms that regulate SP phenotype by changing the overall expression level of ABCG2 since our finding was based on the observations of short-term Akt signaling inhibition. Our findings in MHCC-97L cell line demonstrate the importance of ABCG2 expression and SP cells in the intrinsic drug resistance of HCC, but the molecular mechanism that governs ABCG2 expression in HCC SP cells needs further study. On the other hand, the effect of Akt signaling on ABCG2 indicates that suppressing this oncogenic signaling could help block the drug efflux from MHCC-97L HCC cells, but whether it is a more general mechanism in ABCG2-expressing HCC cells still needs verification. Our study also underlines the use of SP analysis in investigating the functional regulation of cancer cells efflux capacity. As SP analysis is broadly applied in the studies of the ABC transporters (18), it could provide more insights into the efflux-related multiple drug resistance phenomenon. Supplementary material Supplementary Table I, Figures S1 and S2 and Materials and methods can be found at Funding National Key Program for Basic Research of China (973) (2002CB513104); National Natural Science Foundation of China (NSFC/RGC ); Shanghai Science and Technology Program (05JC14028 and ). Acknowledgements Conflict of Interest Statement: None declared. References 1. Hussain,S.A. et al. (2001) Hepatocellular carcinoma. Ann. Oncol., 12, Zhu,A.X. (2006) Systemic therapy of advanced hepatocellular carcinoma: how hopeful should we be? Oncologist, 11, Hayes,J.D. et al. (1990) Molecular mechanisms of drug resistance. Biochem. J., 272, Grude,P. et al. (2002) MDR1 gene expression in hepatocellular carcinoma and the peritumoral liver of patients with and without cirrhosis. Cancer Lett., 186, Borst,P. et al. (2002) Mammalian ABC transporters in health and disease. Annu. Rev. Biochem., 71, Doyle,L.A. et al. (1998) A multidrug resistance transporter from human MCF-7 breast cancer cells. Proc. Natl Acad. Sci. USA, 95, Sarkadi,B. et al. (2004) ABCG2 a transporter for all seasons. FEBS Lett., 567, Vander Borght,S. et al. (2006) Breast cancer resistance protein (BCRP/ ABCG2) is expressed by progenitor cells/reactive ductules and hepatocytes and its expression pattern is influenced by disease etiology and species type: possible functional consequences. J. Histochem. Cytochem., 54, Zen,Y. et al. (2007) Histological and culture studies with respect to ABCG2 expression support the existence of a cancer cell hierarchy in human hepatocellular carcinoma. Am. J. Pathol., 170, Hirschmann-Jax,C. et al. (2004) A distinct side population of cells with high drug efflux capacity in human tumor cells. Proc. Natl Acad. Sci. USA, 101, Challen,G.A. et al. (2006) A side order of stem cells: the SP phenotype. Stem Cells, 24, Goodell,M.A. et al. (1996) Isolation and functional properties of murine hematopoietic stem cells that are replicating in vivo. J. Exp. Med., 183,

9 Side population and intrinsic ABCG2 expression in HCC 13. Wulf,G.G. et al. (2001) A leukemic stem cell with intrinsic drug efflux capacity in acute myeloid leukemia. Blood, 98, Kondo,T. et al. (2004) Persistence of a small subpopulation of cancer stem-like cells in the C6 glioma cell line. Proc. Natl Acad. Sci. USA, 101, Patrawala,L. et al. (2005) Side population is enriched in tumorigenic, stemlike cancer cells, whereas ABCG2þ and ABCG2- cancer cells are similarly tumorigenic. Cancer Res., 65, Zhou,S. et al. (2001) The ABC transporter Bcrp1/ABCG2 is expressed in a wide variety of stem cells and is a molecular determinant of the sidepopulation phenotype. Nat. Med., 7, Scharenberg,C.W. et al. (2002) The ABCG2 transporter is an efficient Hoechst efflux pump and is preferentially expressed by immature human hematopoietic progenitors. Blood, 99, Lemos,C. et al. (2008) Drug transporters: recent advances concerning BCRP and tyrosine kinase inhibitors. Br. J. Cancer, 98, Haraguchi,N. et al. (2006) Characterization of a side population of cancer cells from human gastrointestinal system. Stem Cells, 24, Chiba,T. et al. (2006) Side population purified from hepatocellular carcinoma cells harbors cancer stem cell-like properties. Hepatology, 44, Dean,M. et al. (2005) Tumour stem cells and drug resistance. Nat. Rev. Cancer, 5, Mogi,M. et al. (2003) Akt signaling regulates side population cell phenotype via Bcrp1 translocation. J. Biol. Chem., 278, Opel,D. et al. (2007) Activation of Akt predicts poor outcome in neuroblastoma. Cancer Res., 67, Yin,S. et al. (2007) CD133 positive hepatocellular carcinoma cells possess high capacity for tumorigenicity. Int. J. Cancer, 120, Krishan,A. et al. (1986) Flow cytometric monitoring of cellular anthracycline accumulation in murine leukemic cells. Cancer Res., 46, Frank,N.Y. et al. (2005) ABCB5-mediated doxorubicin transport and chemoresistance in human malignant melanoma. Cancer Res., 65, Lichtenauer,U.D. et al. (2008) Side population does not define stem celllike cancer cells in the adrenocortical carcinoma cell line NCI h295r. Endocrinology, 149, Platet,N. et al. (2007) Fluctuation of the SP/non-SP phenotype in the C6 glioma cell line. FEBS Lett., 581, Burkert,J. et al. (2007) Side populations of gastrointestinal cancers are not enriched in stem cells. J. Pathol., 214, Vivanco,I. et al. (2002) The phosphatidylinositol 3-kinase AKT pathway in human cancer. Nat. Rev. Cancer, 2, Kipp,H. et al. (2002) Trafficking of canalicular ABC transporters in hepatocytes. Annu. Rev. Physiol., 64, Received April 1, 2008; revised September 12, 2008; accepted September 19,