Senescence, an inherent tumor suppressive mechanism, is a critical determinant for chemotherapy. In the present study, we show that the monocarboxylate transporter 2 (MCT2) protein was tumor-selectively expressed in human colorectal malignancies and knockdown of MCT2 induces mitochondrial dysfunction, cell-cycle arrest, and senescence without additional cellular stress in colorectal cancer cell lines. Moreover, the reactive oxygen species (ROS) scavenger, N-acetylcysteine, blocked MCT2 knockdown-induced growth arrest and cellular senescence, indicating a pivotal role of ROS in this pathway. Dramatic induction of mitochondrial superoxide generation and decrease in ATP production was observed, indicating that mitochondrial dysfunction is the major mechanism underlying MCT2 knockdown-induced ROS generation. Senescence-associated DNA damage was also evident from the increase in promyelocytic leukemia bodies, γH2AX foci, and SAHF. Conversely, overexpression of MCT2 prevented doxorubicin-induced ROS accumulation (P = 0.0002) and cell growth inhibition (P = 0.001). MCT2 knockdown suppressed KRAS mutant colorectal tumor growth in vivo. In addition, MCT2 knockdown and cytostatic drug combination further enhanced the antitumor effect. These findings support the use of MCT2 as a promising target for inhibition of colorectal cancer. Mol Cancer Ther; 11(11); 2342–51. ©2012 AACR.
Tumors are usually limited in terms of oxygen availability (1) and adapt to hypoxia by uncoupling their glycolytic metabolism from aerobic respiration (2). Persistence of glycolysis, the Warburg effect, is a typical characteristic of advanced cancers. Glycolysis, which produces only 2 ATPs for a glucose molecule, is considered a less effective pathway than aerobic respiration that generates 38 ATP molecules. However, lactic acid released by glucose-consuming hypoxic tumors (2) is consumed as a predominant source of oxidative metabolism of tumor cells (3). The gatekeepers of this metabolic process are monocarboxylate transporters (MCT; MCT1, 2, 3, and 4), which transport monocarboxylates, including pyruvate and lactate (4).
The MCT family comprises 14 members, among which only the first 4 (MCT1–4) catalyze the proton-linked transport of metabolically important monocarboxylates, such as lactate, pyruvate, and ketone bodies (5). MCT2 displays 10-fold higher affinity for monocarboxylates than the MCT1 and MCT4 uptake mechanisms (4). MCT3 is uniquely expressed in the retinal pigment epithelium. MCT1 and MCT4, but not MCT2, have been shown to interact specifically with CD147, which supports MCT expression on the cell surface. MCT1 is present in almost all tissues, whereas MCT2 is expressed in fewer tissue types, suggesting a unique functional role of this protein (6). MCT2 displays strong cytoplasmic expression, but no membrane expression in cancer (7). Experimental evidence of the presence of MCT2 in the mitochondrial membrane indicates a role in the mitochondrial import of pyruvate following lactic acid oxidation (4). Healthy colonocytes derive 60% to 70% of their energy supply from short-chain fatty acids, particularly butyrate, which is transported across the luminal membrane of the colonic epithelium via MCT1. Therefore, inhibition of MCT1 can retard tumor growth through blocking the energy supply (3). Pinheiro and colleagues reported increased expression of MCTs 1, 2, and 4 in colorectal carcinomas (8). However, earlier analysis of healthy colonic tissues and carcinomas revealed a significant decline in MCT1 protein expression during transition from normality to malignancy (9).
Irreversible cell-cycle arrest or senescence is an inherent tumor suppressive mechanism and is a critical determinant for chemotherapy. We show here that MCT2 is selectively inhibited in DNA damage-induced cellular senescence. Moreover, protein levels of MCT2, but not MCT1, are significantly increased in colorectal carcinomas, compared with those of normal tissues. However, the effects and precise mechanisms of MCT2 inhibition in colorectal cancer remain unknown at present. 5-Fluorouracil (5-FU) is a widely used anticancer drug to treat solid tumors, including colorectal cancers. Currently, there is no effective treatment of KRAS mutated metastatic colorectal cancer, which does not respond to 5-FU combined with irinotecan or oxaliplatin chemotherapy (6). We wondered whether modulation of MCT2 might enhance sensitivity of colorectal cancer to 5-FU in KRAS mutant colorectal cancer cell lines.
In the present study, we investigated the effects of MCT2 knockdown in regulating 5-FU sensitivity of colorectal cancer. Our results showed that increased levels of reactive oxygen species (ROS) mainly mediate MCT2 knockdown, causing an increase in mitochondrial dysfunction and senescence-associated nuclear markers, including promyelocytic leukemia (PML) bodies, γH2AX, and SAHF (10). The antioxidant N-acetylcysteine (NAC), a ROS inhibitor, effectively prevented the cellular changes induced by knockdown of MCT2, supporting the involvement of ROS in this pathway. Specifically, we investigated the effects of MCT2 knockdown alone or in combination with 5-FU on human colorectal cancer xenografted in mice. The selective expression of MCT2 (but not MCT1) protein in human primary colorectal tumors, but not normal tissues, further suggests that targeting MCT2 represents a promising strategy to enhance therapeutic efficacy.
Materials and Methods
Cell culture and reagents
Human colon carcinoma (LoVo, HT29, HCT8, HCT116, SW480, and DLD1) and gastric carcinoma (MKN45 and MKN74) cells were grown in RPMI-1640 medium (Gibco Life Science) supplemented with 10% FBS, 1 mmol/L Na2CO3, 2 mmol/L L-glutamine, and penicillin-streptomycin. Cells were cultured at 37°C in a humidified 5% CO2 environment. Following informed consent and in accordance with the appropriate Institutional Review Boards, tumor specimens were obtained from patients undergoing surgery at the Samsung Medical Center. The 21-nucleotide-long siRNAs targeting MCT1, MCT2, and negative control siRNA (siC) were purchased from Dharmacon. The full-length MCT2 open reading frame was obtained from LoVo mRNA using a reverse transcription-PCR (RT-PCR)-based cloning technique, and inserted into the pEGFP plasmid (Clontech). The level of ectopic MCT2 expression in stable cell lines was analyzed by immunoblotting using an anti-GFP antibody (Santa Cruz). Cells were transfected with siRNA or plasmids using Effectene (Qiagen) or an Amaxa electroporation system (Amaxa), according to the manufacturers' instructions.
RNA interference and transfection
Cells (2 × 105 cells per 60 mm dishes) were transfected with 20 nM siRNAs (Dharmacon, Lafayette, CO) using Effectene transfection reagents (Qiagen) according to the manufacturer's instructions and were used for immunoblot analysis 48 hours after transfection. Sequences of the siRNAs used were control nontargeting siRNA (5′-UAGCGACUAAACACAUCAA-3′), MCT1-targeted siRNA (siMCT1) (5′- CCAAGGCAGGGAAAGAUAAGUCUAA-3′) and MCT2-targeted siRNA (siMCT2) (5′-GGAUUUAACUGGAGAAUAU-3′).
Total cellular RNA was extracted using RNeasy Mini Kit (Qiagen) and treated with DNase I (Qiagen). One microgram of RNA was converted to cDNA using Omniscript RT Kit (Qiagen). The primer sequences designed from the coding region of human MCT2 cDNA are as follows: forward, 5′- AGGATTAATTGCAAACTCCA -3′, and reverse, 5′- CCGAATGTTTAGATTTGCTC-3′. The PCR conditions were as follows: 25 cycles of 95°C for 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds, followed by a final incubation at 72°C for 10 minutes.
Senescence-associated β-galactosidase staining
Cells were seeded into 60-mm dishes in RPMI-1640 culture medium and transfected with siRNA (20 nmol/L), and senescence-associated β-galactosidase (SA-β-Gal) staining was conducted as previously described (6). Senescence was scored based on the percentage of the population that exhibited a SA-β-Gal activity, and the results were photographed under phase contrast microscopy.
Western blot analyses
Total cell extracts were obtained using lysis buffer containing 50 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 0.1% SDS, 1% NP40, and 1X protease inhibitors (Roche Applied Science), and protein concentration was determined using the micro-BCA protein reagent (Pierce). Primary antibodies against the following proteins were used: MCT1 (Sigma, AV43841, 1:1,000), MCT2 (Santa Cruz, H-40, sc-50322, 1:500), p21 (Santa Cruz, C-19, sc-397, 1:1,000), p27 (Santa Cruz, C-19, sc-528, 1:1,000), Rb (BD Pharmingen, G3-245, 554136, 1:1,000), hypophosphorylated Rb (BD Pharmingen, G99-549, 554164, 1:1,000), E2F-1 (Santa Cruz, KH95, sc-251, 1:1,000), cyclin A (Santa Cruz, H-432, sc-751, 1:1,000), Cdc2 (Santa Cruz, 17, sc-54, 1:1,000), GFP (Santa Cruz, B-2, sc-9996, 1:1,000), and β-actin (Sigma, AC-15, A5441, 1:5,000).
Assays for cell proliferation and colony formation assay
To measure cell growth in vitro, cells were grown in RPMI-1640 medium with 10% FBS at 37°C. Cells were seeded (1 × 105 cells per well) in a 6-well plate (Nunc) and incubated at 37°C for 1 day (24 hours). Cells were then transfected with siRNA for an additional 24 hours. Cells were then treated with 0.25% trypsin–EDTA solution (2.5 g/L of trypsin, 0.38 g/L of EDTA; Invitrogen), stained with 0.4% trypan blue solution (Sigma-Aldrich, Inc.), and counted using a hemocytometer (Hausser Scientific). The results were expressed as percentage cell proliferation, using the number of living cells incubated with PBS as a 100% reference. For anchorage-dependent colony formation, we used DLD1, HCT8, and LoVo cells. In brief, 2 × 102 cells per well were seeded in 6-well plates (Nunc) and treated with 0.03 μg/mL 5-FU for 3 days. Triplicate cultures of each cell type were maintained at 37°C for 14 days in an atmosphere of 5% CO2, with fresh medium being added after 7 days. Cells were stained with 0.1% (weight/vol) crystal violet.
Colonies, defined as groups of cells containing a minimum of 50 cells, were counted under an inverted phase contrast microscope. The percentage relative cell proliferation was expressed as (number of colonies from treated cells/number of colonies from controls) × 100. The assay was repeated 3 times with duplicate samples.
For cell-cycle analysis, cells were washed twice with ice-cold PBS, and then fixed in 2 mL of 70% ethanol. The fixed cells were centrifuged at 200 × g for 10 minutes, and pellets were washed twice with PBS. Cells were then incubated concurrently with 40 μg/mL propidium iodide (Sigma) and 100 μg/mL RNase at 37°C for 30 minutes. The percentages of cells in different phases of the cell cycle were measured with a FAC-Star flow cytometer (BD Sciences) and analyzed using Becton-Dickinson software (CellQuest, BD Sciences).
Determination of mitochondrial mass, mitochondrial membrane potential, ROS level, and ATP concentration
LoVo cells (2 × 105 cells per well) were seeded in 6-well plates and incubated at 37°C for 1 day (24 hours). Cells were then transfected with siRNA (20 nmol/L) for an additional 5 days. To measure mitochondrial mass, Mitotracker Red (M7512; Invitrogen) was used. Cells were incubated for 5 minutes with 1 μmol/L Mitotracker Red and the intensity of labeling was measured by fluorescence-activated cell sorting (FACS). To assess mitochondrial membrane potential, JC-1 (Molecular Probes) was used; cells were incubated with 10 μg/mL JC-1 for 10 minutes and washed with Hank's Buffered Salt Solution (Gibco). To measure intracellular production of ROS, we used 2 different fluorogenic probes, DCFH-DA and MitoSox (Molecular Probes). DCFH-DA reacts quantitatively with intracellular radicals, being converted to a fluorescent product, 2′,7′-dichlorofluorescein (DCF). Hydroethidine assay (11) and MitoSox were used to measure mitochondrial superoxide production. Stained cells were washed, resuspended in PBS, and analyzed using a FACS Calibur flow cytometer (BD Sciences). The ATP concentration was determined using an ATP assay kit (FL-ASC; Sigma) and the ATP/AMP ratio measured as described previously (12); data were normalized to cell number.
Immunofluorescence staining of γH2AX, PML, and SAHF
Cells were fixed in 3.7% (v/v) paraformaldehyde for 15 minutes, washed with PBS, and permeabilized using 0.2% (v/v) Triton X-100 in PBS. After blocking with 3% (w/v) bovine serum albumin (BSA) for 30 minutes, the cells were incubated for 1 hour with either anti-γH2AX (1:200; Upstate Technology) or anti-PML (1:200; Santa Cruz) in 3% (w/v) BSA in PBS. Subsequently, cells were washed 3 times with 3% (w/v) BSA in PBS and incubated with Alexa Fluor 488-conjugated secondary antibody (1:5,000) for 1 hour. 4′,6-diamidino-2-phenylindole was added to stain nuclei.
Tumorigenesis of xenotransplanted human colorectal cancer cells
Male BALB/c nude mice, 4 to 6 weeks old, were obtained from Orient Bio Inc. Mice (n = 5 per cell line per treatment group) were implanted subcutaneously with DLD1 (3.0 × 106 cells) in 100 μL volume using a 26-gauge needle. Each mouse received 2 subcutaneous injections in the bilateral flank for the development of 2 tumors (10 tumors per treatment group). One week after implantation, mice (n = 5 mice per treatment group) were assigned into 4 groups—siControl, 5-FU, siMCT2, or a combination of 5-FU and siMCT2. The mice were treated twice per week with intraperitoneal injection of 30 mg/kg 5-FU in PBS and/or once weekly intratumorally with 1 μg siMCT2 dissolved in effectene reagent. Tumor diameters were serially measured with a digital caliper (Mitutoyo) every 2 to 3 days, and tumor volumes were calculated using the following formula: V = (L × W2)/2, where V = volume (cubic millimeter), L = length (millimeter), and W = width (millimeter). The mice were killed by CO2 inhalation, and the tumors were resected on day 15. Mice were handled at the institute's (Samsung Medical Center, Seoul, Korea) animal facility, and all treatments were in accordance with institutional guidelines.
Data presented in graphs represent means ± standard deviations of values from at least 3 independent measurements. Differences between 2 mean values were analyzed using Student t test (paired 2-sample t test). All P values less than 0.05 were considered to be statistically significant.
MCT2 expression decreases according to DNA damage-induced senescence of cancer cells
Doxorubicin, a DNA damage-inducing drug frequently used to treat various types of solid tumors, triggers premature senescence (13). Low-dose doxorubicin induces growth arrest in human carcinoma cells, as evident from their SA-β-Gal positivity and flat, enlarged morphology (14, 15). In our experiments, MCT2 expression was reduced in response to doxorubicin treatment to an increasing extent with the progression of senescence (Fig. 1A). To establish whether MCT2 is directly involved in drug-induced senescence of colorectal cancer cells, we further examined the effects of MCT2 knockdown using specific siRNA (siMCT2). The growth of several cancer cells was specifically inhibited by siMCT2 (Fig. 1B), but not siMCT1 (Supplementary Fig. S1), in a concentration-dependent manner. Analysis of transfected cells revealed a siMCT2-specific decrease in MCT2 mRNA (Fig. 1C), indicative of successful knockdown. This cell growth inhibition was dose-dependent. As shown in Fig. 1D, MCT2 overexpression effectively rescued cells from siMCT2-induced senescence, clearly indicating that these biologic changes are mediated by MCT2.
ROS are involved in MCT2-modulated cell-cycle arrest
Irreversible growth arrest is associated with cell-cycle inhibitors, including p53, pRb, and the cyclin-dependent kinase inhibitors, p21, p27, and p16 (16–18). Accordingly, we investigated the effects of MCT2 knockdown on the levels of cell cycle proteins in KRAS mutated LoVo (KRASG13D) cells. MCT2 knockdown led to increased levels of the G1 arrest-inducing protein, p27, and decreased cdc2 and E2F1 levels and Rb phosphorylation (Fig. 2A). Next, we attempted to determine whether MCT2 overexpression suppresses cell cycle arrest and premature senescence in DNA damage-induced colorectal carcinoma cells. As shown in Fig. 2, doxorubicin-treated colorectal carcinoma cells overexpressing MCT2 exhibited decreased G1 arrest (Fig. 2B) and improved survival (Fig. 2C), compared with doxorubicin-treated control LoVo cells.
ROS accumulation has been reported to induce senescence (10, 19). Accordingly, we investigated whether MCT2 knockdown induces ROS generation. A fluorescent marker of cellular oxidant production, DCFH-DA (20), was used to measure the intracellular levels of ROS in LoVo cells. MCT2 knockdown resulted in dramatic ROS accumulation comparing with MCT1 knockdown (Fig. 3A and Supplementary Fig. S2). Rescue of MCT2 levels via overexpression restored siMCT2-induced ROS to levels comparable to those in controls (P < 0.0001), indicating that accumulation of ROS is directly dependent on the MCT2 level (Fig. 3B, gray). Inhibition of ROS accumulation was previously shown to protect against senescence (21). Thus, we used the ROS scavenger, NAC (22), to determine whether accumulation of ROS plays a role in MCT2 knockdown-induced senescence. As shown in Fig. 3C, preincubation with 10 mmol/L NAC prevented senescence in MCT2 knockdown cells, indicating that senescence is induced as a consequence of ROS accumulation.
MCT2 knockdown induces senescence-associated phenotypes in mitochondria and nucleus
To ascertain whether MCT2 knockdown-induced senescence is a general phenomenon, we evaluated the senescence-inducing potential of siMCT2 in several colorectal cancer cell lines (Fig. 4A). All the cancer cell lines examined, including those of the colon (HCT8, HCT116, HT29, LoVo, and SW480) and stomach (MKN45 and MKN74), displayed cellular enlargement and flattening as well as positivity for SA-β-Gal staining, following knockdown of MCT2. Moreover, depletion of MCT2, but not MCT1, resulted in mitochondrial dysfunction (23), as evident from increased mitochondrial mass (Fig. 4B, a and Supplementary Fig. S3), mitochondrial membrane potential (Fig. 4B, b), elevated mitochondrial ROS production (Fig. 4B, c), and dramatic decrease in the ATP level (Fig. 4B, d). Cells additionally exhibited senescence-associated nuclear properties, including elevation of PML bodies, γH2AX, and SAHF (Fig. 4C).
Autophagy is often induced as a survival pathway to tolerate metabolic stress. To determine whether autophagy is triggered during siMCT2-induced senescence, we evaluated autophagolysosome formation and GFP-LC3 translocation, in addition to SA-β-Gal staining. Accumulation of autophagic vacuoles was detected via fluorescence microscopic observation of GFP-LC3 translocation and Cyto-ID-stained autophagolysosomes. MCT2 knockdown cells presented green punctate structures, indicative of autophagy induction (Fig. 5). To further establish whether autophagy has a protective or toxic effect on siMCT2-transfected LoVo cells, we analyzed the outcomes of autophagy inhibition. Addition of various autophagy inhibitors (3-MA, bafilomycin, and chloroquine) led to the efficient blockage of siMCT2-induced autophagosome formation, but had no significant effects on senescence or survival of siMCT2-transfected cells.
Combined antitumor effects of MCT2 knockdown and 5-FU on colorectal cancer cells in vitro
In view of the above results, we speculated that MCT2 knockdown may enhance the antitumor effect of 5-FU in colon cancer. An anchorage-dependent colony formation assay was conducted to assess antitumor activity. Knockdown of MCT2 to 0.03 μg/mL 5-FU led to significant enhancement of inhibition of tumor cell colony formation in DLD1, HCT8, and LoVo cells (DLD1, 0.03 μg/mL 5-FU vs. 0.03 μg/mL 5-FU + siMCT2, P = 0.0002; HCT8, 0.03 μg/mL 5-FU vs. 0.03 μg/mL 5-FU + siMCT2, P = 0.0002; LoVo, 0.03 μg/mL 5-FU vs. 0.03 μg/mL 5-FU + siMCT2, P = 0.0004; Fig. 6). Clearly, a combination of 0.03 μg/mL 5-FU and siMCT2 has a greater inhibitory effect on cell growth, compared with 5-FU or siMCT2 as single agents. Our findings suggest that siMCT2 sensitizes cancer cells to 5-FU.
Effect of MCT2 on growth of colorectal tumors in vivo
For unknown reason, we failed to make xenotransplant tumors in mice with LoVo (KRASG13D) cells. Therefore, to determine whether MCT2 knockdown also has an antitumor effect in vivo, we implanted DLD1 (KRASG13D) tumors in mice and assigned the animals to 4 groups (n = 5 mice per treatment group), specifically, siControl, siMCT2, 5-FU, and a combination of 5-FU and siMCT2 (Fig. 7A). Knockdown of MCT2 significantly reduced tumor volume, compared with siControl (P = 0.0264). A combination of 5-FU and siMCT2 induced a more significant reduction in tumor volume, compared with 5-FU (P = 0.0118) or siMCT2 (P = 0.0199) xenografts (Fig. 7B). Thus, it appears that MCT2 knockdown suppresses tumorigenicity and enhances the antitumor effect of 5-FU in colorectal cancer in vivo. MCT1 has additionally been implicated in tumor growth in cervix squamous carcinoma (3). Accordingly, we examined the levels of MCT1 and MCT2 in human colon and gastric cancer specimens (Fig. 7). Tumor-specific increases in MCT2, but not MCT1, were observed using Western blot analysis.
In this study, we have shown that knockdown of MCT2 significantly suppresses the growth of cancer cells and enhances the antitumor activity of 5-FU in colorectal cancer cells. In a mouse model, growth of human colorectal xenograft tumors was significantly inhibited upon knockdown of MCT2, which was further suppressed in combination with 5-FU. To our knowledge, this is the first study to report a potential role of MCT2 as a molecular target in colorectal cancer. Previous studies of MCT2 expression in colorectal tumor samples have focused on the use of immunohistochemistry, and these studies have noted changes in the cellular location of MCT2 protein in tumor samples. Differently from MCT1 or MCT4, the immunostaining pattern of MCT2 was cytoplasmic (7, 8). Experimental evidence of the presence of MCT2 in the mitochondrial membrane indicates its role in the mitochondria (4).
Permanent growth arrest or senescence is considered an important determinant of treatment outcomes in cancer therapy (24–27). Our experiments showed that MCT2 knockdown alone induces a senescence-like phenotype in a variety of cancer cell lines, including those established from the colon (HCT8, HCT116, HT29, DLD1, and LoVo) and stomach (MKN45 and MKN74). MCT2 knockdown showed a senescence-associated mitochondrial dysfunction (23), including increases in mitochondrial mass and ROS production and decreases in the ATP level and mitochondrial membrane potential. These cells also displayed other nuclear phenotypes, including increased PML bodies (6), DNA damage-associated γH2AX foci (7), and SAHF (8).
Because MCT2 downregulation induces ROS accumulation in LoVo cells, we hypothesized that ROS is a critical mediator of MCT2 knockdown-induced senescence. ROS is a critical mediator in the induction of senescence (20, 28–30). The antioxidant, NAC, blocked ROS accumulation in response to MCT2 knockdown and protected cells from irreversible growth arrest. Therefore, we show here for the first time that ROS accumulation in response to MCT2 knockdown is responsible for causing mitochondrial dysfunction, cell-cycle arrest, and senescence.
Cellular senescence or irreversible growth arrest is associated with induction of tumor suppressors, such as p53, pRb, p16, p21, and p27 (16–18). These tumor suppressors act through triggering ROS accumulation, which, in turn, induces permanent growth arrest/senescence (21). Therefore, induction of cell-cycle inhibitory tumor suppressors and accumulation of ROS may both contribute to irreversible growth arrest/senescence induced by MCT2 knockdown. Moreover, we found that cells displaying increased MCT2 expression are resistant to growth arrest and able to proliferate in the presence of the DNA-damaging agent, doxorubicin, that normally induces senescence. MCT1 has been previously implicated in colon (8) and cervical cancer (3). However, this is the first study showing that MCT2 is selectively overexpressed in colorectal tumors, but not MCT1. This finding is particularly significant, and supports the theory that MCT2 plays a role in escape from senescence-associated inherent tumor suppressor mechanism in colorectal cancer progression in vivo.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: W.K. Kang, C. Park
Development of methodology: I. Lee, W.K. Kang, C. Park
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Lee, S.J. Lee, W.K. Kang, C. Park
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): I. Lee, W.K. Kang, C. Park
Writing, review, and/or revision of the manuscript: W.K. Kang, C. Park
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W.K. Kang, C. Park
Study supervision: W.K. Kang, C. Park
The authors thank Prof. Kyeong Sook Choi for advice and support on autophagy experiment.
This work was supported by grants from IN-SUNG Foundation (CA98641) to W.K. Kang and National Research Foundation of Korea (NRF-2011-0016973) to C. Park.