NDRG1 inhibition sensitizes osteosarcoma cells to combretastatin A-4 through targeting autophagy

Combretastatin A-4 (CA-4), a tubulin-depolymerizing agent, shows promising antitumor efficacy and has been under several clinical trials in solid tumors for 10 years. Autophagy has an important pro-survival role in cancer therapy, thus targeting autophagy may improve the efficacy of antitumor agents. N-myc downstream-regulated gene 1 (NDRG1) is a significant stress regulatory gene, which mediates cell survival and chemoresistance. Here we reported that CA-4 could induce cell-protective autophagy, and combination treatment of CA-4 and autophagy inhibitor chloroquine (CQ) exerted synergistic cytotoxic effect on human osteosarcoma (OS) cells. Meanwhile, CA-4 or CQ could increase the expression of NDRG1 independently. We further performed mechanistic study to explore how CA-4 and CQ regulate the expression of NDRG1. Using luciferase reporter assay, we found that CA-4 transcriptionally upregulated NDRG1 expression, whereas CQ triggered colocalization of NDRG1 and lysosome, which subsequently prevented lysosome-dependent degradation of NDRG1. Further, we showed that knockdown of NDRG1 caused the defect of lysosomal function, which accumulated LC3-positive autophagosomes by decreasing their fusion with lysosomes. Moreover, NDRG1 inhibition increased apoptosis in response to combination treatment with CA-4 and CQ. Taken together, our study revealed abrogation of NDRG1 expression sensitizes OS cells to CA-4 by suppression of autophagosome–lysosome fusion. These results provide clues for developing more effective cancer therapeutic strategies by the concomitant treatment with CA-4 and clinical available autophagy inhibitors.

Autophagy is an evolutionarily conserved, homeostatic process that components of the cell are degraded to maintain essential activity and viability as a response to numerous stimuli. 1 Autophagy begins with the formation of doublemembrane autophagic vesicles (AVs), known as autophagosomes, which engulf damaged or superfluous proteins and organelles. The autophagosomes subsequently fuses with lysosomes form the autolysosomes (signal-membrane AVs), where the components inside are degraded and recycle. Because of autophagy major role in cell survival during unfavorable conditions, targeting autophagy may be a reasonable anticancer strategy that improves the efficacy of many standard of care agents. Consistent with this viewpoint, growing evidence shows that autophagy inhibitors like chloroquine (CQ) or 3-methyladenine (3-MA) sensitize cancer cells to chemotherapy treatments like DNA-damage agent doxorubicin, 2 DNA alkylating agent cisplatin, 3 microtubuletargeting agent vincristine, 4 anti-angiogenic agent bevacizumab 5 and tyrosine kinase receptor inhibitor imatinib. 6 Hence, understanding how autophagic machinery regulates chemotherapy sensitivity is crucial for cancer therapy.
Combretastatin A-4 (CA-4), a tubulin-depolymerizing agent, shows a great effect in antitumor therapy and has entered clinical trials of solid tumors over 10 years. CA-4 phosphate (CA-4P) is a water-soluble CA-4 prodrug. CA-4 has a high affinity for tubulin, and destabilizes the tubulin polymers of the cytoskeleton, resulting in morphological changes. These changes increase vascular permeability and disrupt tumor blood flow. 7,8 Anti-vascular effects in vivo are seen within minutes of drug administration and rapidly lead to extensive ischemic necrosis in areas that are often resistant to conventional anticancer treatments. 9,10 Recently, increasing evidence has implicated that suppression of autophagy has been suggested to potentially enhance the therapeutic efficacy of CA-4. 11,12 Nevertheless, whether disrupting autophagy would augment the anticancer activity of CA-4 in osteosarcoma (OS) cells is still unknown and needs further clarification.
The N-myc downregulated gene 1 (NDRG1) is a member of the NDRG family, which belongs to the α/β hydrolase superfamily, and overexpressed in several types of human carcinomas. 13 Most intensive studies indicated that the function of NDRG1 is associated with inhibiting cancer metastasis and progression in cancer of brain, breast, colon, rectum, esophagus, pancreas and prostate. [14][15][16] Paradoxically, it has been suggested to promote vascular invasion, metastasis and poor prognosis in cancers of the kidney, liver, mouth, skin and uterine cervix. 17,18 Collectively, NDRG1 has an important role of promoting or inhibiting in cancer patients depending upon the tumor species, histological type and differentiation status of human malignancies. 19 NDRG1 is also recognized as a significant stress response gene and is regulated by a wide range of stress stimuli, such as hypoxia, homocysteine, nickel, androgens, calcium and iron depletion, and chemotherapy. 20 Recently, studies have been suggested that NDRG1 is involved in modulating sensitivity and resistance of cancer cells to chemotherapeutic agents. 21,22 Weiler et al. 23 identified that the mTOR targets NDRG1 as a key determinant of resistance toward alkylating chemotherapy, driven by hypoxia, irradiation, corticosteroids and chronic alkylating agents. Jung et al. 24 also found that hypoxia-and RA-inducible NDRG1 expression is the response for doxorubicin and RA resistance, and the selective interruption of NDRG1 signaling may prove therapeutically useful in hepatocellular carcinoma cells. Hence, these findings indicated that NDRG1 may be developed as an attractive candidate for targeted therapy.
In this study, we investigate the functional and therapeutic relevance of NDRG1 in the combination treatment of CA-4 and autophagy inhibitor CQ against human OS cells. We demonstrated that CA-4 induced cell-protective autophagy, and the combination of CA-4 and CQ exerts synergistic antitumor effects in human OS cells. Meanwhile, both of CA-4 and CQ upregulated the expression of NDRG1 independently. Furthermore, NDRG1 knockdown sensitized human OS cells to CA-4 and CQ combination treatment. Our findings provide clues for developing more effective cancer therapeutic strategies by the concomitant treatment with CA-4 and clinical available autophagy inhibitors.

CA-4 induces autophagy in human OS cells.
To investigate the effect of CA-4 on autophagy and the role of autophagy in determining the sensitivity of OS cells to this agent, we first examined the activity of autophagy in OS cells treated with CA-4. As shown in Figures 1a and b, OS cells treatment with various concentrations of CA-4 caused a dose-dependent activation of autophagy, as evidenced by a significant increase in the conversion of LC3 from LC3B-I to LC3-II. On autophagic induction, LC3B-I is modified and converted to the phagophore-associated LC3-II through conjugating with the lipid phosphatidylethanolamine. LC3-II levels, thus, serve as an index for the number of autophagosomes. 25 Similar results were observed in other OS cells lines (Supplementary Figure S1A). LC3-II levels were further elevated in the presence of CQ, a lysosome inhibitor that blocks the fusion of autophagosomes and lysosomes and LC3-II degradation, indicating an increase of autophagic flux in CA-4 treated OS cells (Figures 1d and e). To further evaluate autophagy induction by CA-4 treatment, we established stable cell lines expressing GFP-tagged LC3, representing the formation of autophagosomes (Figure 1g). After quantitation analysis of the number of GFP-LC3 puncta, we observed an accumulation of autophgosomes visualized by a sixfold in CA-4 treatment transfected cells (Figure 1h). In addition, we also examined the SQSTM1 protein level, a selective autophagy receptor protein that interacts with ubiquitinated proteins and LC3-II. The level of SQSTM1 slightly reduced following CA-4 incubation in OS cells (Figures 1a, c, d and f). These findings suggest that CA-4 induces autophagy in OS cells.
CA-4 and CQ combination treatment has a synergistic effect in OS cells. To explore whether CA-4-stimulated autophagy is pro-survival or pro-death in OS cells, we used the autophagy inhibitor CQ in combination with this agent. First, we observed a concentration-and time-dependent decrease in cell viability as measured by the CCK8 assay after incubation with a range of CA-4 or CQ at 24, 48 and 72 h (Supplementary Figures S2A and B). We tested the IC 50  CA-4 and CQ upregulates NDRG1 expression by different mechanisms, respectively. NDRG1 has been recently recognized as a significant stress regulatory gene, which is upregulated by a wide range of stress stimuli. 13,26 Nevertheless, it is unknown whether NDRG1 is mediated by CA-4 and CQ in OS cells. First, we examined the effects of CA-4 on expression of in OS cell lines, and results showed CA-4 treatment significantly increased NDRG1 protein levels (Figures 3a and b), and the NDRG1 quantitation included two bands (Figure 3b). We further performed RT-PCR to measure the mRNA level of NDRG1 and found CA-4 treatment induced a significant increase of NDRG1 mRNA level in OS cells (Figure 3c). These results suggested that the regulatory effect of CA-4 on NDRG1 expression may occur at the transcription level. As expected, the full length of the NDRG1 promoter was used to show that CA-4 could concentration-dependently induce luciferase activity (Figure 3d). CQ could disrupt lysosomal structure and function, leading to the inhibition of protein degradation. [27][28][29] This led us to investigate CQ effects on the degradation of NDRG1. We pretreated OS cells with different concentrations of CQ, and CA-4 and CQ combination treatment H Wang et al results showed that level of NDRG1 protein was markedly increased in a dose-dependent manner by immunoblotting assay (Figures 3e and f), and the NDRG1 quantitation included two bands ( Figure 3f). However, pretreatment with CQ did not change NDRG1 mRNA level by RT-RCR assays in OS cells (Supplementary Figure S3A), indicating that CQmediated NDRG1 upregulation did not occur at the transcript level, probably because of inhibition of its degradation. To this end, the stability of NDRG1 protein was examined following treatment of cells with protein synthesis inhibitor cycloheximide (CHX). Cells were treated with CHX in the presence or absence of CQ and results showed that the decrease NDRG1 observed after CHX treatment was partially recovered upon co-treatment with CHX and CQ (Figures 3g and h).
To further confirm these findings, we identified NDRG1 whose distinct accumulation patterns trapped in lysosomes in the CQ-treated cells (Figure 5a). These results suggested that a lysosome-dependent activity of CQ prevents degradation of NDRG1 protein.
NDRG1 knockdown causes the accumulation of autophagosomes. We explored whether there was an association between the function of NDRG1 and autophagy in OS cells. First, we used siRNA to knockdown NDRG1 expression, and overall knockdown efficiency of 80% was confirmed by quantification ( Figure 4a). NDRG1 knockdown increased the number of LC3B-II puncta ( (Figures 4b and c). These results were consistent with the previous study, in which NDRG1 was involved in the regulation of stress-induced, pro-survival autophagic pathway. 30 Furthermore, we examined whether NDRG1 functioned in the late stage of autophagy, tandem mCherry-GFP-LC3 fluorescence assay was used. In this assay, autophagosomes appear as yellow (mCherry and GFP) puncta, whereas autolysosomes appear as red (mCherry) puncta. NDRG1-silencing cells exhibited an accumulation of yellow autophagosomes under basal  (Figures 4f and g). Consistently, fluorescent GFP-LC3 puncta were easily observed in NDRG1 knockdown cells (Figures 4h and j), and did not colocalize with lysosome-associated membrane protein 1 (LAMP1) (Figures 4d and k). Collectively, these data suggested that NDRG1 knockdown leads to defects in lysosomal activity. Our data demonstrated that NDRG1 depletion blocks autophagic flux and promotes accumulation of autophagosomes, suggesting that NDRG1 knockdown may impair lysosomal degradation at a point downstream of fusion. We examined the subcellular localization of NDRG1 and LAMP1, a lysosome marker, and results showed the colocalization between NDRG1 and LAMP1 following treatment by CQ (Figure 5a). Furthermore, we investigated the effects of NDRG1 silencing on LAMP1 by immunoblotting, and the results found level of LAMP1 was reduced in NDRG1 knockdown cells (Figure 5b). Based on these findings, we speculated that NDRG1 may be associated with lysosomes. First, we measured the lysosomal acidification, which is required for the maturation and activation of most lysosomal enzymes. OS cells were stained with a specific lysosome dye LysoTracker Red, which accumulates in acidic cellular organelles, and the cells were treated with starvation conditions. We found, compared with nontarget siRNA control cells, the lysosomal fluorescence intensity was lower in NDRG1 knockdown cells (Figures 5c and d,Supplementary  Figures S5A and B). We further quantified lysosomal pH using a LysoSensor Yellow/Blue dye, and found that lysosomal pH in NDRG1 knockdown cells was increased from pH 4.3 in control cells to 5.4 (Figure 5e). These results suggested that NDRG1 silencing suppresses the lysosomal acidification. Second, we performed functional assays of lysosomal activity of OS cells. Fluorogenic substrate assay was used to measure the enzymatic activity of CTSB, a member of the cysteine cathepsin family, and results showed that CTSB activity was reduced in NDRG1 knockdown cells (Figure 5f). Finally, we utilized epidermal growth factor receptor (EGFR) as endogenous substrate to examine whether NDRG1 silencing affects the general endosomallysosomal clearance. EGFR undergoes endocytosis and degrades in lysosomes. In this assay, OS cells transfected with NDRG1 siRNA were treated with EGF. EGFR downregulation was observed after 15-min treatment with EGF, but NDRG1 inhibition attenuated EGFR clearance. However, there was no obvious change in p-AKT and p-ERK between siControl and siNDRG1 cells (Figures 5g and h). Taken together, these data demonstrated that absence of NDRG1 causes defect in lysosomal activity.

NDRG1 knockdown sensitizes OS cells to combination treatment CA-4 and CQ.
We examined whether the marked increases in NDRG1 expression by the CA-4/CQ regimen had a critical role in the sensitivity of OS cells to apoptosis induced by this combination. Silencing of NDRG1 expression by specific siRNA significantly enhanced the cytotoxicity of the combination of CA-4 and CQ, as compared with those transfected with a non-targeted control siRNA (Figure 6a). We further demonstrated that NDRG1 knockdown significantly increased the CA-4 and CQ-induced apoptosis, as indicated by increases in sub-G0/G1 (Figure 6b), and the amounts of cleaved PARP and cleaved caspase 3 (Figures 6c and d).
Suppression of autophagy and activation of apoptosis by inhibiting NDRG1 in the CA-4 and CQ combination treatment cells suggested a regulatory role for NDRG1 in the cross-talk between autophagy and apoptosis. Importantly, these findings established that suppression of NDRG1 increases sensitivity to CA-4 and CQ combination therapy in OS cells.

Discussion
Autophagy inhibition has been shown to enhance chemosensitivity and antitumor efficacy in multiple tumor cell types, and thus the identification and validation of novel strategies to exploit autophagy for therapeutic advantage against cancer are a significant challenge. We hypothesized that inhibition the autophagy pathway may enhance the antitumor activity of CA-4, and the studies demonstrated that autophagy inhibitor CQ synergistically potentiated the pro-apoptotic effects of CA-4. Furthermore, we also found that both of CA-4 and CQ could increase the expression of NDRG1, a stress response gene that mediates cell survival and chemoresistance. CA-4 transcriptionally upregulated NDRG1 expression, whereas CQ triggered colocalization of NDRG1 and lysosome, which subsequently prevented lysosome-dependent degradation of NDRG1. Silencing NDRG1 suppressed the lysosomal acidification, and impaired function of lysosome endocytic clearance, which leaded to accumulation of autophagosomes. Importantly, inhibition of NDRG1 expression increased apoptosis in response to combination treatment with CA-4 and CQ. Our work demonstrated that NDRG1 linked to chemotherapy sensitivity, and our results may have very important clinical implication. Chemotherapy-induced autophagy is cyto-protective; consequently, inhibition of autophagy is anticipated to sensitize malignancies to therapy. These findings laid the foundation for clinical and experimental applications combining of autophagy inhibitors (such as CQ) with chemotherapy agents. 31-33 CQ, which is a classical anti-malarial and anti-inflammatory drug, inhibits lysosomal acidification and therefore prevents autophagy by blocking autophagosomes fusion and degradation. In cancer treatment, CQ and its derivatives are the few inhibitors of autophagy that are available for use in the clinic, multiple ongoing clinical trials used CQ combination with diverse chemotherapeutic drugs for this purpose. 32,34 The vascular targeting agent CA-4 is a microtubule-depolymerizing agent. CA-4P is a water-soluble prodrug of CA-4, and CA-4P is undergoing Phase II and III clinical trials to evaluate the safety and efficacy in human cancer patients when used in combination with chemotherapy or radiation therapy. [35][36][37] In this study, we showed that a combination of CQ could potentiate antitumor effects of CA-4. This finding is also consistent with recent studies, in which CA-4-elicited autophagic response had a protective role that impeded the eventual cell death, whereas autophagy inhibition improved chemotherapeutic efficacy of CA-4. 11,12 We further demonstrated that combination of CA-4 and CQ in OS cells synergistically activated the caspase cascade-induced apoptosis, as compared with single-agent treatment at the same concentrations. Moreover, the concentration of each drug administration in the combination was almost 50% less than the IC 50 of each CA-4 and CQ combination treatment H Wang et al agent, which favors a tolerability of this combination that will be explored in future clinical trials. NDRG1 is recognized as a significant stress regulatory gene and is induced by a range of stress, including starvation, hypoxia, homocysteine, nickel, androgens, iron depletion and chemotherapy. 30,38 As expected, we demonstrated that CA-4 and CQ treatment could increase the expression of NDRG1, respectively, but with different mechanisms. By using RT-PCR and NDRG1 promoter-driven luciferase reporter, we found CA-4 treatment increased NDRG1 expression at the transcriptional level. As to CQ, our results showed that CQ treatment led to increase of NDRG1 expression and specific colocalization of NDRG1 and lysosomes. Further experiments found that NDRG1 treated by CQ are more rapidly turned over than the absence of CQ using CHX chase experiments. Our mechanistic insight into NDRG1 accumulation by CQ treatment focuses on the lysosome degradation pathway. Interestingly, it appears that the higher band is responding to CQ and CA-4 more markedly than the lower band. The producing of NDRG1 double band has remained controversial. Many studies suggested the higher band is due to NDRG1 phosphorylation. 39,40 Indeed, NDRG1 undergoes phosphorylation at multiple sites, resulting in a phosphorylated isoform that is of a higher molecular mass. 41 However, a study using dephosphorylation assay and Phos-tag SDS/ PAGE assay demonstrated that the higher band was not due to phosphorylation, but rather proteolytic cleavage. 42 In addition, one or more of the NDRG1 bands detected by western blot are due to different NDRG1 primary antibodies or nonspecific binding of either primary or secondary antibodies. 43,44 From the above, it is very difficult to speculate the reason of for one or more the NDRG1 bands followed by CQ and CA-4 treatment. It needs further clarification. Next, we broadened our view to examine the effect of NDRG1 on regulation of autophagy. We observed that loss-of-function of NDRG1 causes an increase in the numbers of autophagosomes, suggesting that NDRG1 silencing may lead to inhibition of subsequent autophagosome process. Our results are in agreement with a recent study, which NDRG1 suppressed iron chelator-mediator accumulation of the autophagic marker LC3-II by activation of the PERK/eIF2α pathway. 30 The reduction of autophagosomes degradation could be caused by either blocking autophagosomes fusion with the lysosomes, or attenuating the enzymatic activity of lysosome hydrolases. 45,46 We found the fusion ability of autophagosomes with lysosomes decreased in NDRG1 knockdown cells. As mentioned above, CQ treatment resulted in NDRG1 colocalized with the lysosomes, therefore we examined whether NDRG1 targeted to lysosomes or alters lysosome function. We found that NDRG1 silencing triggered a reduction in proteolytic activity, and an increase in lysosome pH, indicating that loss of NDRG1 in turn reduced the turnover of autophagosomes. It is noteworthy that poor lysosome activity resulting from defects in lysosomal acidification or maturity had an inhibitory impact on endocytic clearance. EGFR is a typical member of the receptor tyrosine kinase family, which after ligand binding-induced activation is endocytosed and delivered to lysosomes for degradation. 46 As expected, further experiments showed that a marked defect in ligand-induced EGFR degradation was observed in cells that NDRG1 silencing. Receptor ubiquitylation occurs at the cell surface, its major role is to sort internalized receptors to the lumen of the multivesicular body, en route to the lysosome. Following EGF binding, EGFR sorting to lysosomes depends on its kinase domain and its ubiquitination by Cbl proteins, a cytosolic ubiquitin ligase. Ubiquitination of EGFR is key mechanism in EGF-induced receptor lysosomal degradation. It is a possible mechanism for NDRG1 markedly and significantly inhibiting level and membrane localization of EGFR in a recent study. 47 NDRG1 activation has frequently been found to promote cancer cell chemoresistance. The molecular mechanism(s) for NDRG1 on chemosensitivity of cancer cells remains unknown. The results of this study imply that autophagy mediated by NDRG1 deficiency via lysosome function impairment may contribute to the sensitivity of cancer cells to chemotherapy drugs. We demonstrated that OS cells transfected with NDRG1 siRNA were significantly higher sensitive to the combination of CQ and CA-4 than those transfected with a non-targeted control siRNA. In contrast to our findings, Sahni et al. 30 reported that NDRG1 upregulation treated by iron chelator-mediator increased the induction of apoptosis in pancreatic cancer cells. This discrepancy may be due to the ability of NDRG1 to regulate autophagy after incubation with iron chelators, which would decrease recycling of nutrients to facilitate apoptotic cell death, as well as differences in the cell types, that is, pancreatic cancer cells versus OS cells. Notably, our previous results showed that NDRG1 expression was increased in OS and this elevation was correlated with tumor progression and prognosis, 48 suggesting that NDRG1 could be considered as a promising therapeutic approach in OS. Therefore, it can be inferred that a combination of NDRG1 inhibition with chemotherapy agents will be used as a useful approach in OS treatment.
In summary, our current studies reveal that CA-4 treatment triggers autophagy, and CA-4 and autophagy inhibitor CQ have a synergistic activity against OS cells (Figure 7). In addition, both of CA-4 and CQ upregulated the expression of NDRG1. We also found that the deficiency of autophagy by NDRG1 silencing enhanced the antitumor effect of the combination by impairing lysosome function. Our findings revealed abrogation of NDRG1 function sensitizes OS cells to CA-4 and CQ combination treatment. These results provide clues for developing more effective cancer therapeutic strategies by the concomitant treatment with CA-4 and clinical available autophagy inhibitors.   Cell apoptosis assay. Apoptosis was evaluated using: (1) flow cytometry via propidum iodide (PI) analysis of sub-G0/G1 DNA content as previous described. 51 (2) Immunoblotting assay of the cleaved-PAPR and cleaved caspases.
Immunofluorescence staining and confocal microscopy. Cells were plated on slips, for different experimental conditions, fixed in 4% paraformaldehyde for 15 min at room temperature, and permeablized with 0.1% Triton X-100 (Sigma, St. Louis, MO, USA, 158127). After blocking with 5% BSA (Sangon Biotech Co., Ltd (Shanghai, China), C508113) in PBS for 30 min, cells were incubated with the following primary antibodies: rabbit anti-NDRG1 (1 : 200) and rabbit anti-LAMP1 (1 : 200) in blocking buffer for at 4°C overnight. The next day, cells were washed and incubated with secondary antibody. Images were obtained using Leica TCS SP5 confocal laser scanning microscope (Leica, Wetzlar, Germany).
Immunoblotting. Cells were lysed in RIPA buffer containing protease and a phosphatase inhibitor cocktail from Roche (Indianapolis, IN, USA). A total of 30 μg protein lysates were separated by 6-12% SDS-PAGE and transferred to PVDF membrane from Bio-Rad (Hercules, CA, USA). The membranes were incubated with primary antibody at 4°C overnight, followed by incubation with peroxidaseconjugated secondary antibody for 1 h at room temperature. The protein signals were detected by ECL from Millipore (Billerica, MA, USA).
Transmission electron microscopy. Cells were treated as indicated and fixed with 2.5% glutaraldehyde solution (Sigma, St. Louis, MO, USA, G5882) at 4°C overnight, and post-fixed in 1% buffered osmium tetroxide for 1.5 h at room temperature. The fixed cells were then dehydrated, embedded and stained with uranyl acetate. Representative areas were chosen for ultrathin sectioning and examined by transmission electron microscopy (FEI Tecnai G 2 12, Eindhoven, The Netherlands).
Measurement of lysosome pH. LysoTracker Red DND-99 staining was used to measure the pH of acidic organelles which become more fluorescent in acidic environments, as according to previous protocols. 52 Briefly, cells from different treatments were incubated with 50 nM of LysoTracker Red DND-99 in the dark for 30 min at 37°C, and the samples were analyzed by fluorescence microscope or flow cytometry. For lysosome, pH quantification was performed with LysoSensor Yellow/Blue DND-160. Cells were seeded in 96-well plates and labeled with 2 μM LysoSensor for 45 min at 37°C in regular medium, then washed in PBS. The labeled cells were treated for 10 min with 10 μM monensin (Sigma, M5273) and 10 M nigericin (Sigma, N7143) in 25 mM 2-(N-morpholino) ethane sulfonic acid (MES) calibration buffer (5 mM NaCl, 115 mM KCl and 1.2 mM MgSO 4 ), which was adjusted with pH from 3.5 to 6.0. The samples were measured with excitation of 335 nm and then the fluorescence emission intensity ratio of 450 nm/520 nm was calculated. The pH value of each sample was determined from the linear standard curve.
Construction of NDRG1 promoter-driven luciferase reporter. Human NDRG1 promoter was inserted in firefly luciferase reporter vector pGL3-Basic by Longqian Biotech (Shanghai, China). Cells were seeded on 24-well plates, and co-transfected with luciferase reporter vectors and Renilla luciferase reporter vector pRL-SV40 (Promega). After 24-h transfection, firefly and Renilla luciferase activities were consecutively measured, according to the dual-luciferase assay manual (Promega, Madison, WI, USA, E1500). The firefly luciferase signal was normalized to the Renilla luciferase signal for each individual analysis.
Statistical analysis. All data were expressed as mean ± standard derivation (S.D.) of three independent experiments, as indicated. Statistical significance was assessed by two-sided Student's t-test or ANOVA. Value of Po0.05 was considered significant and asterisked without correction for multiple statistical tests.

Conflict of Interest
The authors declare no conflict of interest.