- Email: [email protected]
Differential regulation of NM23-H1 under hypoxic and serum starvation conditions in metastatic cancer cells and its implication in EMT
European Journal of Cell Biology 96 (2017) 164–171
Contents lists available at ScienceDirect
European Journal of Cell Biology journal homepage: www.elsevier.com/locate/ejcb
Short communication
Differential regulation of NM23-H1 under hypoxic and serum starvation conditions in metastatic cancer cells and its implication in EMT Reyaz Ur Rasool a,b,1 , Debasis Nayak a,b,1 , Souneek Chakraborty a,b,1 , Vijay Lakshmi Jamwal c , Vidushi Mahajan a,c , Archana Katoch a,b , Mir Mohd Faheem b , Zainab Iqra b , Hina Amin b , Sumit G. Gandhi c,b , Anindya Goswami a,b,∗ a
Academy of Scientific & Innovative Research (AcSIR), CSIR-Indian Institute of Integrative Medicine, Jammu 180001, India Cancer Pharmacology Division, CSIR-Indian Institute of Integrative Medicine, Jammu 180001, India c Plant Biotechnology and System Biology Division, CSIR-Indian Institute of Integrative Medicine, Jammu 180001, India b
a r t i c l e
i n f o
Article history: Received 1 October 2016 Received in revised form 4 December 2016 Accepted 26 January 2017 Keywords: NM23-H1 EMT Hypoxia E-cadherin Panc-1 Serum-starvation TME
a b s t r a c t Multiple stresses are prevalent inside the tumor microenvironment rendering tumor growth, neighboring invasion and metastasis of the cancer cells to distant organs. NM23-H1 is the first metastasis suppressor gene identified and known to be implicated as an important regulator of stress-induced metastasis. Herein, we demonstrated that prototypical NM23-H1 expression diminished during hypoxia and serum starvation in Panc-1/MDA-MB-231 cells, but converse invasion patterns were obtained in these two diverse stresses. Supportingly, a compelling discrete difference in mRNA and protein levels of NM23-H1 was achieved in hypoxia as well as serum starvation. Knockdown of NM23-H1 activates EMT whereas the similar effects are subdued in serum starvation where NM23-H1 down-modulation prompted Ecadherin upregulation. Stable NM23-H1 expression augmented E-cadherin levels along with retardation in invadopodea formation and invasion. In hypoxia/serum starvation excess NM23-H1 effectively modulated the Twist1 promoter activity. Thus, differential regulation of NM23-H1 may corroborate/abrogate EMT depending on the nature of stress, tumor microenvironment and cellular context. © 2017 Elsevier GmbH. All rights reserved.
1. Introduction Metastasis of the tumor cells to distant organs account for more than 90% of the all cancer deaths worldwide (Gupta and Massagu, 2006; Steeg, 2006). The phenomenon of metastasis involves a series of critical events including epithelial-mesenchymal transition (EMT) that ultimately leads to malignancy (Reymond et al., 2013; Stoletov et al., 2010). The cells undergoing metastasis disseminate from the primary tumor and colonize at distant organs to form secondary tumors through angiogenesis (Reymond et al., 2013). In the vicinity of tumor microenvironment, accumulating stress factors including hypoxia, growth factor deprivation, acidosis, pH alteration etc. further trigger active metastasis. Alteration in the expression of many pro and antimetastatic genes takes place
∗ Corresponding author at: CSIR-Indian Institute of Integrative Medicine, Canal Road, Jammu 180001, India. E-mail address: [email protected] (A. Goswami). 1 These authors have contributed equally. http://dx.doi.org/10.1016/j.ejcb.2017.01.008 0171-9335/© 2017 Elsevier GmbH. All rights reserved.
at the molecular level (due to stress) that further aggravates the metastatic cascade and advancement of the disease. Neutralization of these built-up tumor-associated stresses have been proven efficacious to abrogate tumor growth and metastasis (Luo et al., 2006). Hence, better understanding of the process, the mechanisms involved, and identification of the molecular targets are essential prerequisites for the development of novel therapeutic strategies. Non-metastatic gene 23 (NM-23), a nucleoside diphosphate kinase, is the first metastasis suppressor identified due to its reduced expression in metastatic murine melanoma (K-1735 TK) cell lines and an elevated level in low metastatic cell lines (Steeg et al., 1993; Kim et al., 2009). The NM23 gene is located on chromosome 17q 21 that encodes for an 17 kDa protein with 152 amino acids. There are about ten NM23 isoforms characterized in humans so far, of which NM23-H1 is the most studied because of its negative correlation with the metastatic potential of human breast, pancreatic, colorectal, ovarian and hepatocellular carcinomas (Prabhu et al., 2012; Palmieri et al., 2006; Lacombe and Boissan, 2013). NM23 proteins belong to the family of nucleoside diphosphate kinases (NDPK); that catalyze the transfer of
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
165
Fig. 1. Hypoxia and serum starvation down-modulate NM23-H1 and regulate cancer cell invasion. (A) MDA-MB-231 and Panc-1 cells were exposed to hypoxic/normoxic conditions for different time points, as indicated, and the expression of NM23-H1, HIF-1␣, AKT and p-AKT was checked through western blotting. (B) MDA-MB-231 and Panc-1 cells were grown in medium with or without serum for the indicated time points and the expression of NM23-H1, AKT and p-AKT was observed with there specific antibodies. Beta actin was used as the loading control. (C, D) Boyden chamber assay showing the effect of serum starvation and hypoxia on MDA-MB-231 and Panc-1 cell invasion. Data are mean ± S.D. of three similar experiments. (*P < 0.05).
␥-phosphate from nucleoside triphosphates to nucleoside diphosphates through the formation of a high energy phospho-histidine intermediate. These proteins are involved in maintaining intracellular homeostasis of nucleotides and various cellular processes such as differentiation, development, metastatic dissemination or cilia functions (Boissan et al., 2009). The members of the NM23/NDPK family interact directly with the members of dynamin superfamily and, thus, maintain high local GTP concentrations and promote dynamin-dependent membrane remodeling (Boissan et al., 2014). Ample evidences suggest that ectopic overexpression of NM23-H1 retards the metastatic potential of highly aggressive cancer cell lines along with the suppression of pro-metastatic genes without affecting the primary tumor growth (Lee et al., 2009a). A study by Leone et al. demonstrated that exogenous expression of NM23-H1 in highly metastatic K-1735 TK murine melanoma cells resulted in reduced incidence of primary tumor formation, significant reductions in tumor metastatic potential independent of tumor cell growth, and altered responses to the cytokine transforming growth factor (Leone et al., 1991). Another study by the same authors reported that upon transfection of human NM23-H1 cDNA into the human MDA-MB-435 breast carcinoma cells as well as its subsequent subcutaneous injection into the mammary fat pad of nude mice reduced tumor metastatic potential in vivo accompanied by stumpy colonization in soft agar colony formation assay in vitro (Leone et al., 1993). Conversely, reduced expression of NM23-H1 in primary colorectal carcinomas is a predictor of lung and lymph node metastasis (Berney et al., 1999; Zeng et al., 1994; MacDonald et al., 1995; Tokunaga
et al., 1993). A study on transgenic mice lacking mouse NM23 (NM23-M1) in mouse models of hepatocellular carcinoma demonstrates that the lack of NM23-M1 expression promotes metastasis in the SV40 animal model of liver carcinogenesis (Boissan et al., 2005). Silencing of NM23-H1 expression in human hepatoma and colon carcinoma cells resulted in disruption of cell–cell adhesion mediated by E-cadherin, -catenin nuclear translocation, and Tcell factor/lymphoid-enhancing factor-1 transactivation along with increase in cell scattering, motility, invasion and invadopodia formation, upregulating several matrix metalloproteinases (Boissan et al., 2010). NM23-H1 can be modulated through chemical molecules/biological agents to achieve therapeutic effects against cancer metastasis. One such agent is medroxyprogesteron acetate (MPA) known to elevate NM23 expression by binding to the glucocorticoid response element at its promoter. High dose of MPA led to the suppression of anchorage-independent growth and pulmonary metastasis after injection of MDA-MB-231 breast cancer cells in immunocompromised mice in vivo (Marshall et al., 2010). However, many scientific evidences claim that NM23 plays a dual role in cancer progression; while its overexpression has been observed in many primary tumors at early stages, its expression is reduced in highly metastatic, advanced tumors. Hence, it remains an enigma how NM23-H1 is regulated inside the tumor microenvironment and how it regulates various metastatic genes to gain control over tumor progression. In this study, we report for the first time how NM23-H1 behaves differently in response to stress factors such as hypoxia and growth factor deprivation, which are obvious inside a tumor microenviron-
166
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
Fig. 2. Decrease in NM23-H1 by hypoxia and serum starvation has differential effects on EMT induction. (A) Western blot analysis of E-cadherin, NM23-H1 and Twist1 after MDA-MB-231 and Panc-1 were grown in serum starved conditions for the given time. (B) Expressional analysis of E-cadherin, NM23-H1 and Twist1 at transcriptional level after MDA-MB-231 and Panc-1 were grown in serum starved conditions. (C) E-cadherin, NM23-H1 and Twist1 expression was assessed through immunoblotting, after MDA-MB-231 and Panc-1 were exposed to hypoxic (6 h) conditions for the given time point. (D) Expressional analysis of E-cadherin, NM23-H1 and Twist1 at transcriptional level after MDA-MB-231 and Panc-1 were grown in hypoxic situation. (E) MDA-MB-231 and Panc-1 were grown in presence or absence of TGF- and the expression of E-cadherin, NM23-H1 and Twist1 were analyzed through immunoblotting. Beta actin was used as the loading control. (F) Panc-1 cells were grown in presence or absence of TGF- and the mRNA expression was quantified by RT-PCR analysis using beta actin as a house keeping gene. Data represented as mean ± S.D. of three independent experiments. (*P < 0.05, **P < 0.01).
ment. We have also investigated its regulation at the transcriptional level in response to the above mentioned stress conditions, its effect on EMT/metastasis associated genes and its implications on the invasion ability of two aggressive human cancer cell lines of diverse tissue origin, i.e., Panc-1, pancreatic adenocarcinoma, and MDA-MB-231, metastatic breast cancer.
2. Results and discussion NM23-H1, the metastasis-associated gene, encodes 17-kDa nucleoside diphosphate kinase. Although NM23-H1 protein plays an important part in regulating the progression of cancer (Steeg et al., 1993), any functional relationship between this protein and different stresses like genotoxic, hypoxic, serum starvation or endo-
plasmic reticulum remains to be elucidated. The ability of cancer cells to adapt to the extreme micro-environment for the development to a more advanced tumor is one of the foremost challenges in the improvement of cancer therapeutics. How this cancer cell adaptation is regulated remains majorly obscure. Here, primarily, we sought to examine the effect of hypoxia and serum starvation on NM23-H1 expression in Panc-1 and MDA-MB-231 cells. Notably, recent studies also unveil indirect contribution of NM23H1 gene towards early metastatic cascades (Tong et al., 2015; Steeg et al., 2008). Therefore, we grew the cells in hypoxic conditions (1% O2 ) for different time points and the ceiling HIF1␣ (a protein expressed by cancer cells in hypoxic-like micro-environment) induction was observed at 6 h time point, as confirmed by the western blots (Fig. 1A). The expression of NM23-H1 examined, cor-
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
167
Fig. 3. Differential role of NM23-H1 in regulating EMT. (A and B) NM23-H1 was transiently knock-down by si-RNA mediated method in Panc-1 cells and invasion of cells in hypoxic and serum starved conditions was analyzed by boyden chamber assay, the bar graph represents the number of migrated cells as counted by using image.j software. (C) Western blot analysis of EMT markers (Vimintin and E-cadherin) in NM23-H1 silenced condition during hypoxic and serum starved condition in Panc-1 cells. (D) Immunoblot depecting the expressional changes in Vimintin and E-cadherin in hypoxic condition when NM23-H1 is overexpressed transiently by NM23-H1-FLAG construct. Hypoxic stress was applied for 6 h following transfection. (E) Both Panc-1 and Panc-1/NM23-H1 stable cells were cultured on FITC coated gelatin coverslips and the area of gelatin degradation was analyzed which clearly reflectes the migratory potential of the cells. (F) Boyden chamber assay showing the migratory capability of Panc-1/NM23-H1 cells under hypoxic and serum starvation conditions. The bar graph represents the percentage area of degration in hypoxic and serum starved conditions. Data are mean ± S.D. of three similar experiments (*P < 0.05).
responded negatively to the expression of HIF1␣, thus pointing out HIF1␣ expression inversely correlated with NM23-H1 expression in Panc-1 and MDA-MB-231 cells. Further, PI3K-Akt activation is known to play a master regulatory role in hypoxic stress envi-
ronment (Yokoi and Fidler, 2004). Therefore, we determined the p-Akt/total Akt expression in normoxic as well as hypoxic condition and the results showed a sharp increase in p-Akt levels in hypoxic condition but remained unchanged in normoxia (data not
168
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
shown) suggesting Akt activation as a co-parallel event in NM23H1 downregulation in hypoxic environment (Fig. 1A). Tumor cells undergoing serum starvation in vitro partially mimic cells devoid of growth factors or factors responsible for invasion and metastasis of tumor, as is the case in tumor before neo-vascularization takes place. At this stage, the stressed cells struggle to cope up with a changed environment by inducing different signaling pathways and gene expression so that the tumor continues to expand (Hicklin and Ellis, 2005). Here, we sought to identify the changes in NM23-H1 protein levels in serum starvation that could reflect the adapted phenotype of the tumor, a potential therapeutic target. Our results implied that serum depletion significantly decreased the expression of NM23-H1 with a concomitant AKT deactivation (24 h) in both Panc-1 and MDA-MB-231 cells (Fig. 1B). The reduction in p-AKT was due to the deprivation of growth factors. AKT is a key player which positively controls ERK activation (Shi et al., 2012; Zhao et al., 2015), therefore, reduction of NM23-H1 in serum starved condition may be due to the inhibition of ERK pathway, which controls c-MYC, a positive regulator of NM23-H1 (Godfried et al., 2002). This expressional change in NM23-H1 could help us better understand Starved-Fed differences among the cell lines and thereby gain clear insights into the effects of serum starvation on potential cellular responses by the anti-metastatic proteins or the pathways involved. We observed that the invasiveness of Panc-1 and MDA-MB-231 cells through boyden chamber, sharply attenuated in serum starvation, despite a steep decline in NM23-H1 expression suggesting that the motility of these aggressive cells was independent of NM23-H1 expression at least under the current experimental conditions (Fig. 1C, D). Since, tumor cells are prone to become highly aggressive in hypoxic tumor milieu (Chaudary and Hill, 2007; Pennacchietti et al., 2003; Yuen and Diaz, 2014), therefore we examined the invasive ability of Panc-1/MDA-MB-231 cells under these conditions. Contrary to the serum starvation data, the results showed a significant augment in invasion of these cells in hypoxic condition (Fig. 1C, D). Tumor hypoxia/growth factor depletion mediated stress induces epithelial-mesenchymal transition (EMT), which triggers invasion and metastasis, promoting cancer stem cells (CSCs) properties (Pennacchietti et al., 2003; Le et al., 2004). To verify the effect of hypoxia/growth factor depletion-mediated regulation of NM23-H1 on EMT the expression of specific EMT markers including, E-cadherin and Twist1, was analyzed at translational as well as transcriptional levels. The Western blots and RT-PCR results showed a decline in the epithelial marker E-cadherin and steady augment in the mesenchymal marker Twist1 levels respectively, in both Panc-1and MDA-MB-231 cells in hypoxic condition (Fig. 2A, B). On the other hand, in contrast to hypoxia, a drastic elevation of E-cadherin and reduction in Twist1 RNA and protein expression were observed in serum-starved cells (Fig. 2C, D). Intriguingly, metastatic suppressor NM23-H1 mRNA levels augmented significantly in MDA-MB231 cells under both serum starvation and hypoxic condition but steadily downregulated at protein levels in both conditions suggesting a strong post-translational regulation of NM23-H1 protein under stressed condition (Fig. 2C). A recent report demonstrated a link between NM23-H1 and TGF-induced EMT (Zhao et al., 2013) and similarly we found that TGF-induced EMT conferred a robust increase in Twist1 expression and a reduction of NM23-H1 expression both in protein and mRNA level (Fig. 2E, F). Since NM23-H1 expression altered in two different stressed conditions, hypoxia and serum starvation, we sought to examine the invasive behavior of aggressive cells following siRNA-mediated knockdown of endogenous NM23-H1 expression. Nevertheless, while substantial in vitro studies ensured that the elevated NM23H1 protein levels inhibited cell motility/invasion (Parhar et al., 1995; Suzuki et al., 2004). However, its downregulation is associated with increased cell motility (Fig. 3A, B). The knockdown of
NM23-H1 in MDA-MB-231 cells did not alter the invasion capability in hypoxia compared to normoxia as quantified by boyden chamber assay (Fig. 3A, middle row, 3B). In contrast, as serum starvation affects cell invasion, we found a mild to moderate increase in cell invasion in cells treated with siNM23-H1 compared to scramble siRNA treated cells (Fig. 3A, lower row, B). Further, to study the correlation of silencing as well as over-expression of NM23H1 with EMT under serum starvation and hypoxia, we performed siRNA-mediated silencing of NM23-H1 in Panc-1 cells and consequent Western blotting; the results depicted a clear up-regulation of EMT markers including vimentin and Twist1 in hypoxic as well as normoxic condition with reduction of E-Cadherin expression under hypoxia in siNM23-H1 cells compared to scramble siRNA cells. However, in serum starvation, knockdown of NM23-H1 triggered robust increase of E-cadherin expression accompanied by diminished vimentin and Twist1 expression (Fig. 3C). The difference in E-cadherin levels may reflect two distinct pathways activation occurring through a common mediator NM23-H1. Moreover, transient over-expression of NM23-H1 in Panc-1 cells unveiled a moderate reversal of EMT, as evident from the subdued and increased expresson of vimentin and E-cadherin respectively (Fig. 3D). Since down modulation of NM23-H1 inducted survival pathways (as shown in Fig. 1), we investigated whether stable overexpression of NM23-H1 could effect the cell invasion in hypoxia and serum starvation conditions. In order to do that, we generated Panc-1/NM23-H1 stable cell lines and examined the effects of hypoxia/serum starvation on matrix gelatin degradation as well as boyden chamber invasion assay. The results showed that the matrix degradation was significantly abrogated in Panc1/NM23-H1 cells compared to Panc-1 cells in both presence and absence of serum condition. However, hypoxia induced an increase in the matrix degradation (Fig. 3E). In contrast, the Panc-1/NM23H1 cells showed a similar invasive property in both hypoxia as well as normoxia condition. The same result was observed in presence and absence of serum (Fig. 3F). The data suggest that NM23-H1 overexpression maintains the therapeutic potential for abrogating cancer cell motility under hypoxia/serum-starved condition. The reciprocal relationship between TME borne stresses and metastasis suppressor NM23-H1 regulation may confer important cues to understand the hypoxia driven NM23-H1 mediated mechanisms responsible for EMT. The mechanism of how hypoxia modulates the expression of NM23-H1 is current research in our lab. The results obtained so far in serum-starved condition depicted that down regulation of NM23-H1 does not result in upregulation of EMT markers or increase in invasion capabilities of aggressive cells. But the down-modulation of p-AKT in 24 h serum starvation suggested an inhibition of the major growth factors-noncanonical signals responsible for activation of EMT. In order to answer the reason for the surprised down-modulation of NM23-H1, we focused our analysis on the effect of ERK/c-MYC signaling on NM23-H1 (Fournier et al., 2003; Dexheimer et al., 2009). Here, we found that in serum starvation, diminished AKT activation led to abrogation of ERK stimulation (Zhao et al., 2015) and reduction of the expression of a major ERK downstream effector, (Godfried et al., 2002) which is also a positive regulator of NM23-H1. When this situation was rescued by adding growth factors such as VEGF, BFGF and TGF to serum-starved cells, which mimiced the conditions of hypoxic microenvironment, (Lee et al., 2009b) we found an increase of vimentin and reduction of E-Cadherin expression (Fig. 4A). The changes in the expressional pattern of E-cadherin was analyzed through immunocyochemistry. The decrease in the E-cadherin expression by hypoxia was noted to be restored when NM23-H1 was transiently over-expressed in the Panc-1 cells (Fig. 4B). Finally to investigate the control of NM23-H1 on genes involved in EMT at transcriptional level we transfected the Panc-1 cells with
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
169
Fig. 4. Cellular Signalling involving the control of NM23-H1 expression. (A) Immunoblot analysis of p-ERK, ERK, c-MYC, Vimentin, E-cadherin and NM23-H1 in condition where depleation of certain growth factors during serum starvation was replenished by adding VEGF, BFGF & TGF which to some extent mimic the conditions during hypoxia (6 h). (B) Immunocytochemistry results showing the effect of hypoxia in presence or absence of ectopic NM23-H1 on E-cadherin expression in Panc-1 cells. (C) The effect of hypoxia and serum starvation on Twist1 transcription was measured by transfecting Panc-1 cells with Twist1 promoter luciferase construct. (D) The Twist1 transcriptional potential was also analyzed in NM23-H1 oversexpressed Panc-1 as well as NM23-H1 silenced cells. Data are mean ± S.D. of three similar experiments. (E) Schematic diagram depecting the role of NM23-H1 in different stressed situation in invasive cancer cells.
Twist1 promoter-driven luciferase reporter construct along with PGL3 as a suitable control. Within 48 h of transfection Twist1/PGL3, cells were retransfected with either NM23-H1 or empty vector. After exposure to either or serum starvation hypoxia the luciferase activities within the cells were analyzed by using dual glow luciferase assay system. The vector transfected cells showed slightly more luciferase activity under hypoxia as compared to normoxia, but upon transfection with NM23-H1 the luciferase activity decreased in both normoxia as well as hypoxic condition (Fig. 4C). In serum-starved condition the vector transfected cells showed less luciferase activity than the cells grown in 10% serum; as observed in hypoxic condition, the serum starved cells transfected with NM23H1 showed a significantly decreased transcriptional activation of Twist1 promoter (Fig. 4C). We also checked the Twist1 luciferase activity in NM23-H1 knockdown situation and as expected upon
silencing of the metastatic suppressor gene, the Twist1 promoterdriven luciferase activity was significantly increased in hypoxia than normoxia as compared to scramble siRNA (Fig. 4D). In contrast, in serum-starved condition Twist1 luciferase activity significantly reduced in scramble as well as in NM23-H1 silenced cells, compared to 10% serum condition (Fig. 4D). Hence, the Twist1 promoter study underscored that NM23-H1 played a significant role in controlling the expression of Twist1 gene, associated with EMT progression. Collectively, the comprehensive role of NM23-H in EMT during hypoxia/serum starvation is illustrated in the schematic model (Fig. 4E). To summarize the results of our current investigation unveil that NM23-H1 regulation varies differentially during stress generation (due to hypoxia/serum deprivation). The NM23-H1 may not always be associated with the processes involving cancer cell invasion and
170
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
metastasis; and its function may depend on the cellular context and the nature of the stress. However, excess/ectopic NM23-H1 confers a substantial resistance to EMT under hypoxic/serum starved scenario paving the way forward to better understanding of the molecular mechanisms of tumor metastasis. Further, it cannot be overruled that in the initial stages of the cancer, NM23-H1 expression may be retained by various pro-proliferative pathways- such as through c-MYC in a feed-back loop mechanism to prevent anoikis and promote metastatic dormancy. 3. Materials and methods 3.1. Cell cultures and reagents PANC-1, MDA-MB-231 and MCF-7 cells were purchased from European Collection of Cell Culture (ECACC). MCF-7, Panc-1 and MDA-MB-231 cells were cultured in RPMI1640, DMEM and L15 media containing 10% fetal bovine serum from Invitrogen life technologies (Invitrogen, Carlsbad, CA) with penicillin-G (100 U/ml)/streptomycin (100 mg/ml) at 37 ◦ C in a humidified 5% CO2 incubator under standard conditions respectively. Antibodies for NM23-H1 was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA) and was highly specific for NM23-H1 without any cross reaction with its closely related isomer NM23H2 (Chen et al., 2003). Other antibodies such as HIF-1␣, AKT, Twist1, E-cadherin and Vimentin were purchased from Santa Cruz Biotechnology. Phospho-AKT was obtained from Cell Signalling Technology (Cell Signaling Technologies, MA, USA). -actin, Bradford reagent, HEPES, NP-40, DTT, Sodium orthovandate, PMSF and protease inhibitor cocktail were obtained from Sigma Chemicals Co. (St. Louis, MO, USA).
of the cells was quantified by counting the stained cells (cells per five fields) under an inverted microscope. 3.4. Fluorescent gelatin degradation assay Cross-linked-fluorophore (FITC)-conjugated gelatin matrixcoated coverslips were prepared as described (Amin et al., 2015). Gelatin-coated coverslips were quenched with DMEM containing 10% fetal bovine serum at 37 ◦ C for 60 min prior to plating cells. To assess the formation of podosomes or invadopodia and degradation of FITC-gelatin matrix, Panc-1 cells were cultured on FITC gelatincoated coverslips. After 16 h, cells were treated with/without serum and hypoxic/normoxic condition for indicated time points followed by DAPI mounting and observed under FLoid Cell Imaging Station for determining gelatin degradation. Images were further analyzed for determining the threshold area of degradation with the help of Image-J software. 3.5. Generation of NM23-H1 stably expressing cells Panc-1 cells at early passage was revived and grown for further one passage and the healthy cells were transfected with NM23H1-FLAG plasmid (a kind gift by Emily Hua, NIH, Bethesda) using neon transfection system (MPK5000, Invitrogen, and Carlsbad, CA) as per the manufacturer’s protocol. Once the cells were 50–60% confluent, G418 geneticin treatment was employed for 12 weeks and after every 2 days media was changed and fresh G418 was added; the initial concentration of G418 was 600 g/ml and was increased upto 1000 g/ml. At the end of the treatment regimen the cells were subjected to western blotting to check the overexpression status of NM23-H1.
3.2. Immunoblotting Immunoblotting experiments were performed according to the standard procedure as described in Rah et al. (2012). Briefly, following treatments with indicated stress inducing agents, cells were lysed in lysis buffer. The cell lysates were centrifuged at 12,000 rpm for 10 min at 4 ◦ C. Protein concentration was determined by the standard Bradford method. Equal amount (20–30 mg) of protein from each sample was subjected to SDS-PAGE and proteins were transferred to PVDF membranes (PAL), blocked with 5.0% (w/v) non-fat milk in PBS contains 0.1% Tween-20 and probed with indicated antibodies for 3 h at room temperature or overnight at 4 ◦ C. Subsequently blots were washed and probed with species specific secondary antibodies coupled to horseradish peroxidase. Immunoreactive proteins were detected by enhanced chemiluminescence reagent (Millipore, Billerica, MA, USA) and exposed to Biomax light film (Eastman Kodak Co., Rochester, NY, USA). 3.3. Matrigel invasion assay The effect of Serum starvation and hypoxia on cell invasion was determined using BD Biocoat Tumor Invasion Assay System (BD Bioscience, Bedford, MA) according to the instruction of the manufacturer. Briefly, Panc-1 and MDA-MB-231 cells (1.2 × 106 ) cells were cultured in the presence of 1% serum and 10% serum for 24 h in the lower chamber and serum free media into the upper chambers/inserts. Cells were allowed to migrate at 37 ◦ C. For hypoxia similarly these cells were cultured in hypoxic chamber for 8 h and consequently for normoxia cells were kept in normal CO2 incubator for 8 h. After the completion of incubation (indicated time points) Matrigel-coated polycarbonate filters were removed, the non-migrating cells were separated from the upper chamber with a cotton swab and the insert were fixed with methanol and stained with 0.1% crystal violet solution. For each replicate (n = 3), invasion
3.6. siRNA knockdown assay Panc-1 cells were seeded in 60 mm cell culture disc (for Western blotting and Real time PCR) or 96 well plate (for reporter assay) and transient transfection were performed with DharmaFECT1transfection reagent or Neon Transfection System (MPK5000, Invitrogen, and Carlsbad, CA) according to manufacturer’s protocol. The siRNA for NM23-H1 was purchased from Sigma Chemicals Co. (St. Louis, MO, USA). 3.7. Luciferase reporter assay A luciferase reporter assay system was used to measure Twist1 transcription activity by using Twist-1 reporter construct (PGL3Twist1-luc), a kind gift from Dr Mien-Chie Hung, MD Anderson cancer centre, Texas. Panc-1 cells were transfected with NM23H1 vector construct and within 48 h the same cells along with suitable controls were transfected with Twist1-luc/PGL3 vector backbone. NM23-H1 was knocked down in Panc-1 cells by the method described above and NM23-H1 stables were taken and subjected to serum starvation and hypoxic condition. The firefly luciferase activity and Renilla luciferase activity were analyzed by two step process using the Dual Glo reagents (E2940, Promega, Madison, USA) according to the manufacturer’s instruction. Quantification was based on three independent experiments. 3.8. Statistical analysis Data were expressed as the mean ± standard error of three independent experiments and analyzed by the student’s t test. A two sided value of *P < 0.05 was considered significant in all cases.
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
Acknowledgements The work was supported by DBT grant GAP-2105 with IIIM publication number IIIM/1984/2016. We thank our Director Dr. R.A. Vishwakarma for encouraging us to complete this work. We would also like to thank Emily Hua, NIH, Bethesda for a kind gift of NM23H1-FLAG plasmid. References Amin, H., et al., 2015. Par-4 dependent modulation of cellular beta-catenin by medicinal plant natural product derivative 3-azido Withaferin A. Mol. Carcinog. 55 (May (5)), 864–881. Berney, C.R., Fisher, R.J., Yang, J.-l., Russell, P.J., Crowe, P.J., 1999. Protein markers in colorectal cancer: predictors of liver metastasis. Ann. Surg. 230, 179. Boissan, M., et al., 2005. Increased lung metastasis in transgenic NM23-Null/SV40 mice with hepatocellular carcinoma. J. Natl. Cancer Inst. 97, 836–845. Boissan, M., et al., 2009. The mammalian Nm23/NDPK family: from metastasis control to cilia movement. Mol. Cell. Biochem. 329, 51–62. Boissan, M., et al., 2010. Implication of metastasis suppressor NM23-H1 in maintaining adherens junctions and limiting the invasive potential of human cancer cells. Cancer Res. 70, 7710–7722. Boissan, M., et al., 2014. Nucleoside diphosphate kinases fuel dynamin superfamily proteins with GTP for membrane remodeling. Science 344, 1510–1515. Chaudary, N., Hill, R.P., 2007. Hypoxia and metastasis. Clin. Cancer Res. 13, 1947–1949. Chen, S.L., et al., 2003. p53 is a regulator of the metastasis suppressor gene Nm23-H1. Mol. Carcinog. 36, 204–214. Dexheimer, T.S., et al., 2009. NM23-H2 may play an indirect role in transcriptional activation of c-myc gene expression but does not cleave the nuclease hypersensitive element III1. Mol. Cancer Ther. 8, 1363–1377. Fournier, H.-N., Albiges-Rizo, C., Block, M.R., 2003. New insights into Nm23 control of cell adhesion and migration. J. Bioenerg. Biomembr. 35, 81–87. Godfried, M.B., et al., 2002. The N-myc and c-myc downstream pathways include the chromosome 17q genes nm23-H1 and nm23-H2. Oncogene 21, 2097–2101. Gupta, G.P., Massagu, J., 2006. Cancer metastasis: building a framework. Cell 127, 679–695. Hicklin, D.J., Ellis, L.M., 2005. Role of the vascular endothelial growth factor pathway in tumor growth and angiogenesis. J. Clin. Oncol. 23, 1011–1027. Kim, H.D., et al., 2009. Regulators affecting the metastasis suppressor activity of Nm23-H1. Mol. Cell. Biochem. 329, 167–173. Lacombe, M.-L., Boissan, M., 2013. Gene Sect. 526 http://AtlasGeneticsOncology. org. Le, Q.-T., Denko, N.C., Giaccia, A.J., 2004. Hypoxic gene expression and metastasis. Cancer Metastasis Rev. 23, 293–310. Lee, J.H., Marshall, J.-C., Steeg, P.S., Horak, C.E., 2009a. Altered gene and protein expression by Nm23-H1 in metastasis suppression. Mol. Cell. Biochem. 329, 141–148. Lee, E.Y., et al., 2009b. Hypoxia enhanced wound healing function of adipose derived stem cells: increase in stem cell proliferation and up-regulation of VEGF and bFGF. Wound Repair Regen. 17, 540–547. Leone, A., et al., 1991. Reduced tumor incidence, metastatic potential, and cytokine responsiveness of nm3-transfected melanoma cells. Cell 65, 25–35. Leone, A., Flatow, U., VanHoutte, K., Steeg, P.S., 1993. Transfection of human nm23-H1 into the human MDA-MB-435 breast carcinoma cell line: effects on tumor metastatic potential, colonization and enzymatic activity. Oncogene 8, 2325–2333.
171
Luo, Y., et al., 2006. Targeting tumor-associated macrophages as a novel strategy against breast cancer. J. Clin. Invest. 116, 2132–2141. MacDonald, N.J., La Rosa, D., Steeg, P.S., 1995. The potential roles of nm23 in cancer metastasis and cellular differentiation. Eur. J. Cancer 31, 1096–1100. Marshall, J.-C., Collins, J., Marino, N., Steeg, P., 2010. The Nm23-H1 metastasis suppressor as a translational target. Eur. J. Cancer 46, 1278–1282. Palmieri, D., Horak, C.E., Lee, J.-H., Halverson, D.O., Steeg, P.S., 2006. Translational approaches using metastasis suppressor genes. J. Bioenerg. Biomembr. 38, 151–161. Parhar, R.S., et al., 1995. Effects of cytokine mediated modulation of nm23 expression on the invasion and metastatic behavior of B16F10 melanoma cells. Int. J. Cancer 60, 204–210. Pennacchietti, S., et al., 2003. Hypoxia promotes invasive growth by transcriptional activation of the met protooncogene. Cancer Cell 3, 347–361. Prabhu, V.V., Siddikuzzaman, S., Grace, V.M., Guruvayoorappan, C., 2012. Targeting tumor metastasis by regulating Nm23 gene expression. Asian Pac. J. Cancer Prev. 13, 3539–3548. Rah, B., Amin, H., Yousuf, K., Khan, S., Jamwal, G., Mukherjee, D., Goswami, A., 2012. A novel MMP-2 inhibitor 3-azido Withaferin A (3-azidoWA) abrogates cancer cell invasion and angiogenesis by modulating extracellular Par-4. PloS One 7, e44039, PMID:22962598 http://dx.doi.org/10.1371/journal.pone.0044039. Reymond, Nicolas, d’Água1, Bárbara Borda, Ridley, Anne J., 2013. Crossing the endothelial barrier during metastasis. Nat. Rev. Cancer 13, 858–870. Shi, Y., et al., 2012. Starvation-induced activation of ATM/Chk2/p53 signaling sensitizes cancer cells to cisplatin. BMC Cancer 12, 1. Steeg, P.S., et al., 1993. Nm23 and breast cancer metastasis. Breast Cancer Res. Treat. 25, 175–187. Steeg, P.S., Horak, C.E., Miller, K.D., 2008. Clinical-translational approaches to the Nm23-H1 metastasis suppressor. Clin. Cancer Res. 14, 5006–5012. Steeg, P.S., 2006. Tumor metastasis: mechanistic insights and clinical challenges. Nat. Med. 12, 895–904. Stoletov, K., et al., 2010. Visualizing extravasation dynamics of metastatic tumor cells. J. Cell Sci. 123, 2332–2341. Suzuki, E., et al., 2004. nm23H1 reduces in vitro cell migration and the liver metastatic potential of colon cancer cells by regulating myosin light chain phosphorylation. Int. J. Cancer 108, 207–211. Tokunaga, Y., et al., 1993. Reduced expression of nm23-H1, but not of nm23-H2, is concordant with the frequency of lymph-node metastasis of human breast cancer. Int. J. Cancer 55, 66–71. Tong, Y., Yung, L.Y., Wong, Y.H., 2015. Metastasis suppressors Nm23H1 and Nm23H2 differentially regulate neoplastic transformation and tumorigenesis. Cancer Lett. 361, 207–217. Yokoi, K., Fidler, I.J., 2004. Hypoxia increases resistance of human pancreatic cancer cells to apoptosis induced by gemcitabine. Clin. Cancer Res. 10, 2299–2306. Yuen, A., Diaz, B., 2014. The impact of hypoxia in pancreatic cancer invasion and metastasis. Hypoxia 2, 91–106. Zeng, Z.S., et al., 1994. High level of Nm23-H1 gene expression is associated with local colorectal cancer progression not with metastases. Br. J. Cancer 70, 1025. Zhao, R., et al., 2013. nm23-H1 is a negative regulator of TGF-beta1-dependent induction of epithelial to mesenchymal transition. Exp. Cell Res. 319, 740–749. Zhao, Q., et al., 2015. Inhibition of c-MYC with involvement of ERK/JNK/MAPK and AKT pathways as a novel mechanism for shikonin and its derivatives in killing leukemia cells. Oncotarget 6, 38934.
Contents lists available at ScienceDirect
European Journal of Cell Biology journal homepage: www.elsevier.com/locate/ejcb
Short communication
Differential regulation of NM23-H1 under hypoxic and serum starvation conditions in metastatic cancer cells and its implication in EMT Reyaz Ur Rasool a,b,1 , Debasis Nayak a,b,1 , Souneek Chakraborty a,b,1 , Vijay Lakshmi Jamwal c , Vidushi Mahajan a,c , Archana Katoch a,b , Mir Mohd Faheem b , Zainab Iqra b , Hina Amin b , Sumit G. Gandhi c,b , Anindya Goswami a,b,∗ a
Academy of Scientific & Innovative Research (AcSIR), CSIR-Indian Institute of Integrative Medicine, Jammu 180001, India Cancer Pharmacology Division, CSIR-Indian Institute of Integrative Medicine, Jammu 180001, India c Plant Biotechnology and System Biology Division, CSIR-Indian Institute of Integrative Medicine, Jammu 180001, India b
a r t i c l e
i n f o
Article history: Received 1 October 2016 Received in revised form 4 December 2016 Accepted 26 January 2017 Keywords: NM23-H1 EMT Hypoxia E-cadherin Panc-1 Serum-starvation TME
a b s t r a c t Multiple stresses are prevalent inside the tumor microenvironment rendering tumor growth, neighboring invasion and metastasis of the cancer cells to distant organs. NM23-H1 is the first metastasis suppressor gene identified and known to be implicated as an important regulator of stress-induced metastasis. Herein, we demonstrated that prototypical NM23-H1 expression diminished during hypoxia and serum starvation in Panc-1/MDA-MB-231 cells, but converse invasion patterns were obtained in these two diverse stresses. Supportingly, a compelling discrete difference in mRNA and protein levels of NM23-H1 was achieved in hypoxia as well as serum starvation. Knockdown of NM23-H1 activates EMT whereas the similar effects are subdued in serum starvation where NM23-H1 down-modulation prompted Ecadherin upregulation. Stable NM23-H1 expression augmented E-cadherin levels along with retardation in invadopodea formation and invasion. In hypoxia/serum starvation excess NM23-H1 effectively modulated the Twist1 promoter activity. Thus, differential regulation of NM23-H1 may corroborate/abrogate EMT depending on the nature of stress, tumor microenvironment and cellular context. © 2017 Elsevier GmbH. All rights reserved.
1. Introduction Metastasis of the tumor cells to distant organs account for more than 90% of the all cancer deaths worldwide (Gupta and Massagu, 2006; Steeg, 2006). The phenomenon of metastasis involves a series of critical events including epithelial-mesenchymal transition (EMT) that ultimately leads to malignancy (Reymond et al., 2013; Stoletov et al., 2010). The cells undergoing metastasis disseminate from the primary tumor and colonize at distant organs to form secondary tumors through angiogenesis (Reymond et al., 2013). In the vicinity of tumor microenvironment, accumulating stress factors including hypoxia, growth factor deprivation, acidosis, pH alteration etc. further trigger active metastasis. Alteration in the expression of many pro and antimetastatic genes takes place
∗ Corresponding author at: CSIR-Indian Institute of Integrative Medicine, Canal Road, Jammu 180001, India. E-mail address: [email protected] (A. Goswami). 1 These authors have contributed equally. http://dx.doi.org/10.1016/j.ejcb.2017.01.008 0171-9335/© 2017 Elsevier GmbH. All rights reserved.
at the molecular level (due to stress) that further aggravates the metastatic cascade and advancement of the disease. Neutralization of these built-up tumor-associated stresses have been proven efficacious to abrogate tumor growth and metastasis (Luo et al., 2006). Hence, better understanding of the process, the mechanisms involved, and identification of the molecular targets are essential prerequisites for the development of novel therapeutic strategies. Non-metastatic gene 23 (NM-23), a nucleoside diphosphate kinase, is the first metastasis suppressor identified due to its reduced expression in metastatic murine melanoma (K-1735 TK) cell lines and an elevated level in low metastatic cell lines (Steeg et al., 1993; Kim et al., 2009). The NM23 gene is located on chromosome 17q 21 that encodes for an 17 kDa protein with 152 amino acids. There are about ten NM23 isoforms characterized in humans so far, of which NM23-H1 is the most studied because of its negative correlation with the metastatic potential of human breast, pancreatic, colorectal, ovarian and hepatocellular carcinomas (Prabhu et al., 2012; Palmieri et al., 2006; Lacombe and Boissan, 2013). NM23 proteins belong to the family of nucleoside diphosphate kinases (NDPK); that catalyze the transfer of
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
165
Fig. 1. Hypoxia and serum starvation down-modulate NM23-H1 and regulate cancer cell invasion. (A) MDA-MB-231 and Panc-1 cells were exposed to hypoxic/normoxic conditions for different time points, as indicated, and the expression of NM23-H1, HIF-1␣, AKT and p-AKT was checked through western blotting. (B) MDA-MB-231 and Panc-1 cells were grown in medium with or without serum for the indicated time points and the expression of NM23-H1, AKT and p-AKT was observed with there specific antibodies. Beta actin was used as the loading control. (C, D) Boyden chamber assay showing the effect of serum starvation and hypoxia on MDA-MB-231 and Panc-1 cell invasion. Data are mean ± S.D. of three similar experiments. (*P < 0.05).
␥-phosphate from nucleoside triphosphates to nucleoside diphosphates through the formation of a high energy phospho-histidine intermediate. These proteins are involved in maintaining intracellular homeostasis of nucleotides and various cellular processes such as differentiation, development, metastatic dissemination or cilia functions (Boissan et al., 2009). The members of the NM23/NDPK family interact directly with the members of dynamin superfamily and, thus, maintain high local GTP concentrations and promote dynamin-dependent membrane remodeling (Boissan et al., 2014). Ample evidences suggest that ectopic overexpression of NM23-H1 retards the metastatic potential of highly aggressive cancer cell lines along with the suppression of pro-metastatic genes without affecting the primary tumor growth (Lee et al., 2009a). A study by Leone et al. demonstrated that exogenous expression of NM23-H1 in highly metastatic K-1735 TK murine melanoma cells resulted in reduced incidence of primary tumor formation, significant reductions in tumor metastatic potential independent of tumor cell growth, and altered responses to the cytokine transforming growth factor (Leone et al., 1991). Another study by the same authors reported that upon transfection of human NM23-H1 cDNA into the human MDA-MB-435 breast carcinoma cells as well as its subsequent subcutaneous injection into the mammary fat pad of nude mice reduced tumor metastatic potential in vivo accompanied by stumpy colonization in soft agar colony formation assay in vitro (Leone et al., 1993). Conversely, reduced expression of NM23-H1 in primary colorectal carcinomas is a predictor of lung and lymph node metastasis (Berney et al., 1999; Zeng et al., 1994; MacDonald et al., 1995; Tokunaga
et al., 1993). A study on transgenic mice lacking mouse NM23 (NM23-M1) in mouse models of hepatocellular carcinoma demonstrates that the lack of NM23-M1 expression promotes metastasis in the SV40 animal model of liver carcinogenesis (Boissan et al., 2005). Silencing of NM23-H1 expression in human hepatoma and colon carcinoma cells resulted in disruption of cell–cell adhesion mediated by E-cadherin, -catenin nuclear translocation, and Tcell factor/lymphoid-enhancing factor-1 transactivation along with increase in cell scattering, motility, invasion and invadopodia formation, upregulating several matrix metalloproteinases (Boissan et al., 2010). NM23-H1 can be modulated through chemical molecules/biological agents to achieve therapeutic effects against cancer metastasis. One such agent is medroxyprogesteron acetate (MPA) known to elevate NM23 expression by binding to the glucocorticoid response element at its promoter. High dose of MPA led to the suppression of anchorage-independent growth and pulmonary metastasis after injection of MDA-MB-231 breast cancer cells in immunocompromised mice in vivo (Marshall et al., 2010). However, many scientific evidences claim that NM23 plays a dual role in cancer progression; while its overexpression has been observed in many primary tumors at early stages, its expression is reduced in highly metastatic, advanced tumors. Hence, it remains an enigma how NM23-H1 is regulated inside the tumor microenvironment and how it regulates various metastatic genes to gain control over tumor progression. In this study, we report for the first time how NM23-H1 behaves differently in response to stress factors such as hypoxia and growth factor deprivation, which are obvious inside a tumor microenviron-
166
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
Fig. 2. Decrease in NM23-H1 by hypoxia and serum starvation has differential effects on EMT induction. (A) Western blot analysis of E-cadherin, NM23-H1 and Twist1 after MDA-MB-231 and Panc-1 were grown in serum starved conditions for the given time. (B) Expressional analysis of E-cadherin, NM23-H1 and Twist1 at transcriptional level after MDA-MB-231 and Panc-1 were grown in serum starved conditions. (C) E-cadherin, NM23-H1 and Twist1 expression was assessed through immunoblotting, after MDA-MB-231 and Panc-1 were exposed to hypoxic (6 h) conditions for the given time point. (D) Expressional analysis of E-cadherin, NM23-H1 and Twist1 at transcriptional level after MDA-MB-231 and Panc-1 were grown in hypoxic situation. (E) MDA-MB-231 and Panc-1 were grown in presence or absence of TGF- and the expression of E-cadherin, NM23-H1 and Twist1 were analyzed through immunoblotting. Beta actin was used as the loading control. (F) Panc-1 cells were grown in presence or absence of TGF- and the mRNA expression was quantified by RT-PCR analysis using beta actin as a house keeping gene. Data represented as mean ± S.D. of three independent experiments. (*P < 0.05, **P < 0.01).
ment. We have also investigated its regulation at the transcriptional level in response to the above mentioned stress conditions, its effect on EMT/metastasis associated genes and its implications on the invasion ability of two aggressive human cancer cell lines of diverse tissue origin, i.e., Panc-1, pancreatic adenocarcinoma, and MDA-MB-231, metastatic breast cancer.
2. Results and discussion NM23-H1, the metastasis-associated gene, encodes 17-kDa nucleoside diphosphate kinase. Although NM23-H1 protein plays an important part in regulating the progression of cancer (Steeg et al., 1993), any functional relationship between this protein and different stresses like genotoxic, hypoxic, serum starvation or endo-
plasmic reticulum remains to be elucidated. The ability of cancer cells to adapt to the extreme micro-environment for the development to a more advanced tumor is one of the foremost challenges in the improvement of cancer therapeutics. How this cancer cell adaptation is regulated remains majorly obscure. Here, primarily, we sought to examine the effect of hypoxia and serum starvation on NM23-H1 expression in Panc-1 and MDA-MB-231 cells. Notably, recent studies also unveil indirect contribution of NM23H1 gene towards early metastatic cascades (Tong et al., 2015; Steeg et al., 2008). Therefore, we grew the cells in hypoxic conditions (1% O2 ) for different time points and the ceiling HIF1␣ (a protein expressed by cancer cells in hypoxic-like micro-environment) induction was observed at 6 h time point, as confirmed by the western blots (Fig. 1A). The expression of NM23-H1 examined, cor-
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
167
Fig. 3. Differential role of NM23-H1 in regulating EMT. (A and B) NM23-H1 was transiently knock-down by si-RNA mediated method in Panc-1 cells and invasion of cells in hypoxic and serum starved conditions was analyzed by boyden chamber assay, the bar graph represents the number of migrated cells as counted by using image.j software. (C) Western blot analysis of EMT markers (Vimintin and E-cadherin) in NM23-H1 silenced condition during hypoxic and serum starved condition in Panc-1 cells. (D) Immunoblot depecting the expressional changes in Vimintin and E-cadherin in hypoxic condition when NM23-H1 is overexpressed transiently by NM23-H1-FLAG construct. Hypoxic stress was applied for 6 h following transfection. (E) Both Panc-1 and Panc-1/NM23-H1 stable cells were cultured on FITC coated gelatin coverslips and the area of gelatin degradation was analyzed which clearly reflectes the migratory potential of the cells. (F) Boyden chamber assay showing the migratory capability of Panc-1/NM23-H1 cells under hypoxic and serum starvation conditions. The bar graph represents the percentage area of degration in hypoxic and serum starved conditions. Data are mean ± S.D. of three similar experiments (*P < 0.05).
responded negatively to the expression of HIF1␣, thus pointing out HIF1␣ expression inversely correlated with NM23-H1 expression in Panc-1 and MDA-MB-231 cells. Further, PI3K-Akt activation is known to play a master regulatory role in hypoxic stress envi-
ronment (Yokoi and Fidler, 2004). Therefore, we determined the p-Akt/total Akt expression in normoxic as well as hypoxic condition and the results showed a sharp increase in p-Akt levels in hypoxic condition but remained unchanged in normoxia (data not
168
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
shown) suggesting Akt activation as a co-parallel event in NM23H1 downregulation in hypoxic environment (Fig. 1A). Tumor cells undergoing serum starvation in vitro partially mimic cells devoid of growth factors or factors responsible for invasion and metastasis of tumor, as is the case in tumor before neo-vascularization takes place. At this stage, the stressed cells struggle to cope up with a changed environment by inducing different signaling pathways and gene expression so that the tumor continues to expand (Hicklin and Ellis, 2005). Here, we sought to identify the changes in NM23-H1 protein levels in serum starvation that could reflect the adapted phenotype of the tumor, a potential therapeutic target. Our results implied that serum depletion significantly decreased the expression of NM23-H1 with a concomitant AKT deactivation (24 h) in both Panc-1 and MDA-MB-231 cells (Fig. 1B). The reduction in p-AKT was due to the deprivation of growth factors. AKT is a key player which positively controls ERK activation (Shi et al., 2012; Zhao et al., 2015), therefore, reduction of NM23-H1 in serum starved condition may be due to the inhibition of ERK pathway, which controls c-MYC, a positive regulator of NM23-H1 (Godfried et al., 2002). This expressional change in NM23-H1 could help us better understand Starved-Fed differences among the cell lines and thereby gain clear insights into the effects of serum starvation on potential cellular responses by the anti-metastatic proteins or the pathways involved. We observed that the invasiveness of Panc-1 and MDA-MB-231 cells through boyden chamber, sharply attenuated in serum starvation, despite a steep decline in NM23-H1 expression suggesting that the motility of these aggressive cells was independent of NM23-H1 expression at least under the current experimental conditions (Fig. 1C, D). Since, tumor cells are prone to become highly aggressive in hypoxic tumor milieu (Chaudary and Hill, 2007; Pennacchietti et al., 2003; Yuen and Diaz, 2014), therefore we examined the invasive ability of Panc-1/MDA-MB-231 cells under these conditions. Contrary to the serum starvation data, the results showed a significant augment in invasion of these cells in hypoxic condition (Fig. 1C, D). Tumor hypoxia/growth factor depletion mediated stress induces epithelial-mesenchymal transition (EMT), which triggers invasion and metastasis, promoting cancer stem cells (CSCs) properties (Pennacchietti et al., 2003; Le et al., 2004). To verify the effect of hypoxia/growth factor depletion-mediated regulation of NM23-H1 on EMT the expression of specific EMT markers including, E-cadherin and Twist1, was analyzed at translational as well as transcriptional levels. The Western blots and RT-PCR results showed a decline in the epithelial marker E-cadherin and steady augment in the mesenchymal marker Twist1 levels respectively, in both Panc-1and MDA-MB-231 cells in hypoxic condition (Fig. 2A, B). On the other hand, in contrast to hypoxia, a drastic elevation of E-cadherin and reduction in Twist1 RNA and protein expression were observed in serum-starved cells (Fig. 2C, D). Intriguingly, metastatic suppressor NM23-H1 mRNA levels augmented significantly in MDA-MB231 cells under both serum starvation and hypoxic condition but steadily downregulated at protein levels in both conditions suggesting a strong post-translational regulation of NM23-H1 protein under stressed condition (Fig. 2C). A recent report demonstrated a link between NM23-H1 and TGF-induced EMT (Zhao et al., 2013) and similarly we found that TGF-induced EMT conferred a robust increase in Twist1 expression and a reduction of NM23-H1 expression both in protein and mRNA level (Fig. 2E, F). Since NM23-H1 expression altered in two different stressed conditions, hypoxia and serum starvation, we sought to examine the invasive behavior of aggressive cells following siRNA-mediated knockdown of endogenous NM23-H1 expression. Nevertheless, while substantial in vitro studies ensured that the elevated NM23H1 protein levels inhibited cell motility/invasion (Parhar et al., 1995; Suzuki et al., 2004). However, its downregulation is associated with increased cell motility (Fig. 3A, B). The knockdown of
NM23-H1 in MDA-MB-231 cells did not alter the invasion capability in hypoxia compared to normoxia as quantified by boyden chamber assay (Fig. 3A, middle row, 3B). In contrast, as serum starvation affects cell invasion, we found a mild to moderate increase in cell invasion in cells treated with siNM23-H1 compared to scramble siRNA treated cells (Fig. 3A, lower row, B). Further, to study the correlation of silencing as well as over-expression of NM23H1 with EMT under serum starvation and hypoxia, we performed siRNA-mediated silencing of NM23-H1 in Panc-1 cells and consequent Western blotting; the results depicted a clear up-regulation of EMT markers including vimentin and Twist1 in hypoxic as well as normoxic condition with reduction of E-Cadherin expression under hypoxia in siNM23-H1 cells compared to scramble siRNA cells. However, in serum starvation, knockdown of NM23-H1 triggered robust increase of E-cadherin expression accompanied by diminished vimentin and Twist1 expression (Fig. 3C). The difference in E-cadherin levels may reflect two distinct pathways activation occurring through a common mediator NM23-H1. Moreover, transient over-expression of NM23-H1 in Panc-1 cells unveiled a moderate reversal of EMT, as evident from the subdued and increased expresson of vimentin and E-cadherin respectively (Fig. 3D). Since down modulation of NM23-H1 inducted survival pathways (as shown in Fig. 1), we investigated whether stable overexpression of NM23-H1 could effect the cell invasion in hypoxia and serum starvation conditions. In order to do that, we generated Panc-1/NM23-H1 stable cell lines and examined the effects of hypoxia/serum starvation on matrix gelatin degradation as well as boyden chamber invasion assay. The results showed that the matrix degradation was significantly abrogated in Panc1/NM23-H1 cells compared to Panc-1 cells in both presence and absence of serum condition. However, hypoxia induced an increase in the matrix degradation (Fig. 3E). In contrast, the Panc-1/NM23H1 cells showed a similar invasive property in both hypoxia as well as normoxia condition. The same result was observed in presence and absence of serum (Fig. 3F). The data suggest that NM23-H1 overexpression maintains the therapeutic potential for abrogating cancer cell motility under hypoxia/serum-starved condition. The reciprocal relationship between TME borne stresses and metastasis suppressor NM23-H1 regulation may confer important cues to understand the hypoxia driven NM23-H1 mediated mechanisms responsible for EMT. The mechanism of how hypoxia modulates the expression of NM23-H1 is current research in our lab. The results obtained so far in serum-starved condition depicted that down regulation of NM23-H1 does not result in upregulation of EMT markers or increase in invasion capabilities of aggressive cells. But the down-modulation of p-AKT in 24 h serum starvation suggested an inhibition of the major growth factors-noncanonical signals responsible for activation of EMT. In order to answer the reason for the surprised down-modulation of NM23-H1, we focused our analysis on the effect of ERK/c-MYC signaling on NM23-H1 (Fournier et al., 2003; Dexheimer et al., 2009). Here, we found that in serum starvation, diminished AKT activation led to abrogation of ERK stimulation (Zhao et al., 2015) and reduction of the expression of a major ERK downstream effector, (Godfried et al., 2002) which is also a positive regulator of NM23-H1. When this situation was rescued by adding growth factors such as VEGF, BFGF and TGF to serum-starved cells, which mimiced the conditions of hypoxic microenvironment, (Lee et al., 2009b) we found an increase of vimentin and reduction of E-Cadherin expression (Fig. 4A). The changes in the expressional pattern of E-cadherin was analyzed through immunocyochemistry. The decrease in the E-cadherin expression by hypoxia was noted to be restored when NM23-H1 was transiently over-expressed in the Panc-1 cells (Fig. 4B). Finally to investigate the control of NM23-H1 on genes involved in EMT at transcriptional level we transfected the Panc-1 cells with
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
169
Fig. 4. Cellular Signalling involving the control of NM23-H1 expression. (A) Immunoblot analysis of p-ERK, ERK, c-MYC, Vimentin, E-cadherin and NM23-H1 in condition where depleation of certain growth factors during serum starvation was replenished by adding VEGF, BFGF & TGF which to some extent mimic the conditions during hypoxia (6 h). (B) Immunocytochemistry results showing the effect of hypoxia in presence or absence of ectopic NM23-H1 on E-cadherin expression in Panc-1 cells. (C) The effect of hypoxia and serum starvation on Twist1 transcription was measured by transfecting Panc-1 cells with Twist1 promoter luciferase construct. (D) The Twist1 transcriptional potential was also analyzed in NM23-H1 oversexpressed Panc-1 as well as NM23-H1 silenced cells. Data are mean ± S.D. of three similar experiments. (E) Schematic diagram depecting the role of NM23-H1 in different stressed situation in invasive cancer cells.
Twist1 promoter-driven luciferase reporter construct along with PGL3 as a suitable control. Within 48 h of transfection Twist1/PGL3, cells were retransfected with either NM23-H1 or empty vector. After exposure to either or serum starvation hypoxia the luciferase activities within the cells were analyzed by using dual glow luciferase assay system. The vector transfected cells showed slightly more luciferase activity under hypoxia as compared to normoxia, but upon transfection with NM23-H1 the luciferase activity decreased in both normoxia as well as hypoxic condition (Fig. 4C). In serum-starved condition the vector transfected cells showed less luciferase activity than the cells grown in 10% serum; as observed in hypoxic condition, the serum starved cells transfected with NM23H1 showed a significantly decreased transcriptional activation of Twist1 promoter (Fig. 4C). We also checked the Twist1 luciferase activity in NM23-H1 knockdown situation and as expected upon
silencing of the metastatic suppressor gene, the Twist1 promoterdriven luciferase activity was significantly increased in hypoxia than normoxia as compared to scramble siRNA (Fig. 4D). In contrast, in serum-starved condition Twist1 luciferase activity significantly reduced in scramble as well as in NM23-H1 silenced cells, compared to 10% serum condition (Fig. 4D). Hence, the Twist1 promoter study underscored that NM23-H1 played a significant role in controlling the expression of Twist1 gene, associated with EMT progression. Collectively, the comprehensive role of NM23-H in EMT during hypoxia/serum starvation is illustrated in the schematic model (Fig. 4E). To summarize the results of our current investigation unveil that NM23-H1 regulation varies differentially during stress generation (due to hypoxia/serum deprivation). The NM23-H1 may not always be associated with the processes involving cancer cell invasion and
170
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
metastasis; and its function may depend on the cellular context and the nature of the stress. However, excess/ectopic NM23-H1 confers a substantial resistance to EMT under hypoxic/serum starved scenario paving the way forward to better understanding of the molecular mechanisms of tumor metastasis. Further, it cannot be overruled that in the initial stages of the cancer, NM23-H1 expression may be retained by various pro-proliferative pathways- such as through c-MYC in a feed-back loop mechanism to prevent anoikis and promote metastatic dormancy. 3. Materials and methods 3.1. Cell cultures and reagents PANC-1, MDA-MB-231 and MCF-7 cells were purchased from European Collection of Cell Culture (ECACC). MCF-7, Panc-1 and MDA-MB-231 cells were cultured in RPMI1640, DMEM and L15 media containing 10% fetal bovine serum from Invitrogen life technologies (Invitrogen, Carlsbad, CA) with penicillin-G (100 U/ml)/streptomycin (100 mg/ml) at 37 ◦ C in a humidified 5% CO2 incubator under standard conditions respectively. Antibodies for NM23-H1 was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA) and was highly specific for NM23-H1 without any cross reaction with its closely related isomer NM23H2 (Chen et al., 2003). Other antibodies such as HIF-1␣, AKT, Twist1, E-cadherin and Vimentin were purchased from Santa Cruz Biotechnology. Phospho-AKT was obtained from Cell Signalling Technology (Cell Signaling Technologies, MA, USA). -actin, Bradford reagent, HEPES, NP-40, DTT, Sodium orthovandate, PMSF and protease inhibitor cocktail were obtained from Sigma Chemicals Co. (St. Louis, MO, USA).
of the cells was quantified by counting the stained cells (cells per five fields) under an inverted microscope. 3.4. Fluorescent gelatin degradation assay Cross-linked-fluorophore (FITC)-conjugated gelatin matrixcoated coverslips were prepared as described (Amin et al., 2015). Gelatin-coated coverslips were quenched with DMEM containing 10% fetal bovine serum at 37 ◦ C for 60 min prior to plating cells. To assess the formation of podosomes or invadopodia and degradation of FITC-gelatin matrix, Panc-1 cells were cultured on FITC gelatincoated coverslips. After 16 h, cells were treated with/without serum and hypoxic/normoxic condition for indicated time points followed by DAPI mounting and observed under FLoid Cell Imaging Station for determining gelatin degradation. Images were further analyzed for determining the threshold area of degradation with the help of Image-J software. 3.5. Generation of NM23-H1 stably expressing cells Panc-1 cells at early passage was revived and grown for further one passage and the healthy cells were transfected with NM23H1-FLAG plasmid (a kind gift by Emily Hua, NIH, Bethesda) using neon transfection system (MPK5000, Invitrogen, and Carlsbad, CA) as per the manufacturer’s protocol. Once the cells were 50–60% confluent, G418 geneticin treatment was employed for 12 weeks and after every 2 days media was changed and fresh G418 was added; the initial concentration of G418 was 600 g/ml and was increased upto 1000 g/ml. At the end of the treatment regimen the cells were subjected to western blotting to check the overexpression status of NM23-H1.
3.2. Immunoblotting Immunoblotting experiments were performed according to the standard procedure as described in Rah et al. (2012). Briefly, following treatments with indicated stress inducing agents, cells were lysed in lysis buffer. The cell lysates were centrifuged at 12,000 rpm for 10 min at 4 ◦ C. Protein concentration was determined by the standard Bradford method. Equal amount (20–30 mg) of protein from each sample was subjected to SDS-PAGE and proteins were transferred to PVDF membranes (PAL), blocked with 5.0% (w/v) non-fat milk in PBS contains 0.1% Tween-20 and probed with indicated antibodies for 3 h at room temperature or overnight at 4 ◦ C. Subsequently blots were washed and probed with species specific secondary antibodies coupled to horseradish peroxidase. Immunoreactive proteins were detected by enhanced chemiluminescence reagent (Millipore, Billerica, MA, USA) and exposed to Biomax light film (Eastman Kodak Co., Rochester, NY, USA). 3.3. Matrigel invasion assay The effect of Serum starvation and hypoxia on cell invasion was determined using BD Biocoat Tumor Invasion Assay System (BD Bioscience, Bedford, MA) according to the instruction of the manufacturer. Briefly, Panc-1 and MDA-MB-231 cells (1.2 × 106 ) cells were cultured in the presence of 1% serum and 10% serum for 24 h in the lower chamber and serum free media into the upper chambers/inserts. Cells were allowed to migrate at 37 ◦ C. For hypoxia similarly these cells were cultured in hypoxic chamber for 8 h and consequently for normoxia cells were kept in normal CO2 incubator for 8 h. After the completion of incubation (indicated time points) Matrigel-coated polycarbonate filters were removed, the non-migrating cells were separated from the upper chamber with a cotton swab and the insert were fixed with methanol and stained with 0.1% crystal violet solution. For each replicate (n = 3), invasion
3.6. siRNA knockdown assay Panc-1 cells were seeded in 60 mm cell culture disc (for Western blotting and Real time PCR) or 96 well plate (for reporter assay) and transient transfection were performed with DharmaFECT1transfection reagent or Neon Transfection System (MPK5000, Invitrogen, and Carlsbad, CA) according to manufacturer’s protocol. The siRNA for NM23-H1 was purchased from Sigma Chemicals Co. (St. Louis, MO, USA). 3.7. Luciferase reporter assay A luciferase reporter assay system was used to measure Twist1 transcription activity by using Twist-1 reporter construct (PGL3Twist1-luc), a kind gift from Dr Mien-Chie Hung, MD Anderson cancer centre, Texas. Panc-1 cells were transfected with NM23H1 vector construct and within 48 h the same cells along with suitable controls were transfected with Twist1-luc/PGL3 vector backbone. NM23-H1 was knocked down in Panc-1 cells by the method described above and NM23-H1 stables were taken and subjected to serum starvation and hypoxic condition. The firefly luciferase activity and Renilla luciferase activity were analyzed by two step process using the Dual Glo reagents (E2940, Promega, Madison, USA) according to the manufacturer’s instruction. Quantification was based on three independent experiments. 3.8. Statistical analysis Data were expressed as the mean ± standard error of three independent experiments and analyzed by the student’s t test. A two sided value of *P < 0.05 was considered significant in all cases.
R.U. Rasool et al. / European Journal of Cell Biology 96 (2017) 164–171
Acknowledgements The work was supported by DBT grant GAP-2105 with IIIM publication number IIIM/1984/2016. We thank our Director Dr. R.A. Vishwakarma for encouraging us to complete this work. We would also like to thank Emily Hua, NIH, Bethesda for a kind gift of NM23H1-FLAG plasmid. References Amin, H., et al., 2015. Par-4 dependent modulation of cellular beta-catenin by medicinal plant natural product derivative 3-azido Withaferin A. Mol. Carcinog. 55 (May (5)), 864–881. Berney, C.R., Fisher, R.J., Yang, J.-l., Russell, P.J., Crowe, P.J., 1999. Protein markers in colorectal cancer: predictors of liver metastasis. Ann. Surg. 230, 179. Boissan, M., et al., 2005. Increased lung metastasis in transgenic NM23-Null/SV40 mice with hepatocellular carcinoma. J. Natl. Cancer Inst. 97, 836–845. Boissan, M., et al., 2009. The mammalian Nm23/NDPK family: from metastasis control to cilia movement. Mol. Cell. Biochem. 329, 51–62. Boissan, M., et al., 2010. Implication of metastasis suppressor NM23-H1 in maintaining adherens junctions and limiting the invasive potential of human cancer cells. Cancer Res. 70, 7710–7722. Boissan, M., et al., 2014. Nucleoside diphosphate kinases fuel dynamin superfamily proteins with GTP for membrane remodeling. Science 344, 1510–1515. Chaudary, N., Hill, R.P., 2007. Hypoxia and metastasis. Clin. Cancer Res. 13, 1947–1949. Chen, S.L., et al., 2003. p53 is a regulator of the metastasis suppressor gene Nm23-H1. Mol. Carcinog. 36, 204–214. Dexheimer, T.S., et al., 2009. NM23-H2 may play an indirect role in transcriptional activation of c-myc gene expression but does not cleave the nuclease hypersensitive element III1. Mol. Cancer Ther. 8, 1363–1377. Fournier, H.-N., Albiges-Rizo, C., Block, M.R., 2003. New insights into Nm23 control of cell adhesion and migration. J. Bioenerg. Biomembr. 35, 81–87. Godfried, M.B., et al., 2002. The N-myc and c-myc downstream pathways include the chromosome 17q genes nm23-H1 and nm23-H2. Oncogene 21, 2097–2101. Gupta, G.P., Massagu, J., 2006. Cancer metastasis: building a framework. Cell 127, 679–695. Hicklin, D.J., Ellis, L.M., 2005. Role of the vascular endothelial growth factor pathway in tumor growth and angiogenesis. J. Clin. Oncol. 23, 1011–1027. Kim, H.D., et al., 2009. Regulators affecting the metastasis suppressor activity of Nm23-H1. Mol. Cell. Biochem. 329, 167–173. Lacombe, M.-L., Boissan, M., 2013. Gene Sect. 526 http://AtlasGeneticsOncology. org. Le, Q.-T., Denko, N.C., Giaccia, A.J., 2004. Hypoxic gene expression and metastasis. Cancer Metastasis Rev. 23, 293–310. Lee, J.H., Marshall, J.-C., Steeg, P.S., Horak, C.E., 2009a. Altered gene and protein expression by Nm23-H1 in metastasis suppression. Mol. Cell. Biochem. 329, 141–148. Lee, E.Y., et al., 2009b. Hypoxia enhanced wound healing function of adipose derived stem cells: increase in stem cell proliferation and up-regulation of VEGF and bFGF. Wound Repair Regen. 17, 540–547. Leone, A., et al., 1991. Reduced tumor incidence, metastatic potential, and cytokine responsiveness of nm3-transfected melanoma cells. Cell 65, 25–35. Leone, A., Flatow, U., VanHoutte, K., Steeg, P.S., 1993. Transfection of human nm23-H1 into the human MDA-MB-435 breast carcinoma cell line: effects on tumor metastatic potential, colonization and enzymatic activity. Oncogene 8, 2325–2333.
171
Luo, Y., et al., 2006. Targeting tumor-associated macrophages as a novel strategy against breast cancer. J. Clin. Invest. 116, 2132–2141. MacDonald, N.J., La Rosa, D., Steeg, P.S., 1995. The potential roles of nm23 in cancer metastasis and cellular differentiation. Eur. J. Cancer 31, 1096–1100. Marshall, J.-C., Collins, J., Marino, N., Steeg, P., 2010. The Nm23-H1 metastasis suppressor as a translational target. Eur. J. Cancer 46, 1278–1282. Palmieri, D., Horak, C.E., Lee, J.-H., Halverson, D.O., Steeg, P.S., 2006. Translational approaches using metastasis suppressor genes. J. Bioenerg. Biomembr. 38, 151–161. Parhar, R.S., et al., 1995. Effects of cytokine mediated modulation of nm23 expression on the invasion and metastatic behavior of B16F10 melanoma cells. Int. J. Cancer 60, 204–210. Pennacchietti, S., et al., 2003. Hypoxia promotes invasive growth by transcriptional activation of the met protooncogene. Cancer Cell 3, 347–361. Prabhu, V.V., Siddikuzzaman, S., Grace, V.M., Guruvayoorappan, C., 2012. Targeting tumor metastasis by regulating Nm23 gene expression. Asian Pac. J. Cancer Prev. 13, 3539–3548. Rah, B., Amin, H., Yousuf, K., Khan, S., Jamwal, G., Mukherjee, D., Goswami, A., 2012. A novel MMP-2 inhibitor 3-azido Withaferin A (3-azidoWA) abrogates cancer cell invasion and angiogenesis by modulating extracellular Par-4. PloS One 7, e44039, PMID:22962598 http://dx.doi.org/10.1371/journal.pone.0044039. Reymond, Nicolas, d’Água1, Bárbara Borda, Ridley, Anne J., 2013. Crossing the endothelial barrier during metastasis. Nat. Rev. Cancer 13, 858–870. Shi, Y., et al., 2012. Starvation-induced activation of ATM/Chk2/p53 signaling sensitizes cancer cells to cisplatin. BMC Cancer 12, 1. Steeg, P.S., et al., 1993. Nm23 and breast cancer metastasis. Breast Cancer Res. Treat. 25, 175–187. Steeg, P.S., Horak, C.E., Miller, K.D., 2008. Clinical-translational approaches to the Nm23-H1 metastasis suppressor. Clin. Cancer Res. 14, 5006–5012. Steeg, P.S., 2006. Tumor metastasis: mechanistic insights and clinical challenges. Nat. Med. 12, 895–904. Stoletov, K., et al., 2010. Visualizing extravasation dynamics of metastatic tumor cells. J. Cell Sci. 123, 2332–2341. Suzuki, E., et al., 2004. nm23H1 reduces in vitro cell migration and the liver metastatic potential of colon cancer cells by regulating myosin light chain phosphorylation. Int. J. Cancer 108, 207–211. Tokunaga, Y., et al., 1993. Reduced expression of nm23-H1, but not of nm23-H2, is concordant with the frequency of lymph-node metastasis of human breast cancer. Int. J. Cancer 55, 66–71. Tong, Y., Yung, L.Y., Wong, Y.H., 2015. Metastasis suppressors Nm23H1 and Nm23H2 differentially regulate neoplastic transformation and tumorigenesis. Cancer Lett. 361, 207–217. Yokoi, K., Fidler, I.J., 2004. Hypoxia increases resistance of human pancreatic cancer cells to apoptosis induced by gemcitabine. Clin. Cancer Res. 10, 2299–2306. Yuen, A., Diaz, B., 2014. The impact of hypoxia in pancreatic cancer invasion and metastasis. Hypoxia 2, 91–106. Zeng, Z.S., et al., 1994. High level of Nm23-H1 gene expression is associated with local colorectal cancer progression not with metastases. Br. J. Cancer 70, 1025. Zhao, R., et al., 2013. nm23-H1 is a negative regulator of TGF-beta1-dependent induction of epithelial to mesenchymal transition. Exp. Cell Res. 319, 740–749. Zhao, Q., et al., 2015. Inhibition of c-MYC with involvement of ERK/JNK/MAPK and AKT pathways as a novel mechanism for shikonin and its derivatives in killing leukemia cells. Oncotarget 6, 38934.