Metformin alleviates nickel-induced autophagy and apoptosis via inhibition of hexokinase-2, activating lipocalin-2, in human bronchial epithelial cells

Autophagy is an intracellular recycling and degradation process for regulating tumor progression, survival and drug resistance. Nickel compounds have been identified as human carcinogens. However, the role of nickel-induced autophagy in lung carcinogenesis has not yet been fully elucidated. In this study, we determined that hexokinase 2 (HK2), which phosphorylates glucose and regulates autophagy, is the key mediator in nickel-induced autophagy in lung bronchial epithelial cells. We attempted to investigate the effects of the antidiabetic drug metformin on HK2 expression and lung cancer chemoprevention. Our results showed that metformin decreases nickel-induced autophagy and activation of apoptosis through inhibition of HK2 gene, protein and activity. Furthermore, we demonstrated that lipocalin 2 (LCN2), which is released by neutrophils at sites of infection and inflammation is involved in HK2-driven autophagy pathway. Knockdown of endogenous HK2 and LCN2 by shRNA reduced nickel-elicited autophagy and apoptosis, illustrating that metabolic alteration and inflammatory action are important in nickel-elicited carcinogenesis. We also determined the association between nickel-induced autophagy and apoptosis. Inhibition of nickel-induced autophagy abolished apoptotic cell death in chloroquine-treated, shLC3 Beas-2B cells and Atg5−/− MFFs. From TGCA database and immunohistochemistry analysis, HK2 and LCN2 expression increased in lung squamous cell carcinoma and their related adjacent normal tissues. Taken together, our results demonstrated that metformin alleviates NiCl2-induced autophagy and apoptosis via HK2-driven LCN2 activation in human bronchial epithelial cells. This novel mechanism provides a strategy for targeting nickel-elicited lung cancer progression, as well as for preventing HK2 cumulative damage triggered by environmental carcinogens.


INTRODUCTION
Autophagy is a highly conserved self-degradative process that packages dysfunctional proteins and organelles into cytoplasmic double-membrane vesicles called autophagosomes. It is activated in response to multiple stimuli in cancer progression, providing a source of nutrients and energy for tumor development during starvation, hypoxia and immune response [1,2]. Nickel is widely distributed, posing occupational and environmental exposure risks. Nickel (II) compounds have been classified as group I human carcinogens by the International Agency Research in Cancer (IARC) of the World Health Organization (WHO). Human exposure to nickel comes from many sources, such as metal industries, rechargeable batteries, electroplating processes and cigarette smoking [3]. Epidemiological studies have indicated that longterm exposure to nickel compounds is the main reason for raised lung and sinus cancer risks among nickel refinery employees [4,5]. The molecular carcinogenic mechanisms of nickel toxicity are thought to involve oxidative stress, DNA damage, epigenetic alteration, chronic inflammation and regulation of gene expression [6,7]. In our previous study, exposure to soluble nickel compound nickel chloride (NiCl 2 ) induced epithelialmesenchymal transition (EMT) in lung epithelial BEAS-2B cells by reactive oxygen species (ROS) generation [8], exhibiting the carcinogenic potential of NiCl 2 . Although the mechanisms of nickel-induced carcinogenesis have been discussed in detail, nickel-activated autophagy has yet to be fully elucidated.
Metabolic alteration is frequently accompanied by rapid differentiation of cells and malignant cells to provide energy. There is a propensity to metabolize glucose to lactic acid by aerobic glycolysis even under sufficient oxygen, a phenomenon known as the Warburg effect [9]. The critical regulator in this frequent cancer phenotype is mitochondrial-bound hexokinase (HK). HK catalyzes the first step of glycolysis, responsible for the conversion of glucose into glucose-6-phosphate. It is well known that HK2, which is found in insulin-sensitive tissues, such as skeletal muscle and adipose tissue, is the major bound HK isoform expressed in cancers [10,11]. High level of HK2 expression is associated with poor overall survival and prognosis in many types of cancers [12,13]. Previous studies have indicated that HK2 facilitates autophagy in response to glucose deprivation, functioning as a molecular switch from glycolysis to autophagy under glucose starvation [14]. However, the manner in which nickel-elicited HK2 contributes to lung carcinogenesis by activating autophagy has yet to be thoroughly investigated.
Lipocalin 2 (LCN2) is a critical inflammatory mediator that is persistently induced during endotoxemia, reflecting the extent of kidney damage and kidney failure [15]. LCN2 is also associated with the pathogenesis of various diseases and is upregulated in many types of cancers. Moreover, it has recently been implicated in multiple cancer tumorigeneses [16,17]. LCN2 also enhance excessive cell autophagy during ischemia/ reperfusion injury [18]. LCN2 deficiency decreases autophagy and inhibits cell proliferation [19]. LCN2 may be an important regulator of tumorigenesis through autophagy and proliferation. In the present study, we investigate the effects of exposure to nickel compounds on LCN2-mediated autophagy.
Metformin is the first-line prescribed drug of choice in the treatment of type 2 diabetes and metabolic syndrome, and has recently emerged as a potential anticancer agent with unanticipated cancer prevention activity. Several epidemiological and clinical studies have found that patients using metformin have decreased cancer incidences, as well as inhibited cancer survival and proliferation, in comparison with those using other antidiabetic medications [20][21][22]. Numerous publications show that anti-carcinogenic effects of metformin are raised by chemical carcinogens or ionizing irradiation in animal models [23,24].
Although it has been established that nickel exposure upregulates HK2 expression, the mechanism of signal integration between nickel-induced HK2 and autophagy in lung tumor progression has not been elucidated. It exhibits anticancer effect of metformin through regulation of glucose metabolism. Consequently, the aim of this study is to investigate the role of metformin in diminishing NiCl 2induced HK2, and associated autophagy and cytotoxicity, in lung bronchial cells. This is the first report of HK2driven inflammatory cytokine LCN2 expression in the promotion of autophagy under nickel exposure. We also clarify the relationship between NiCl 2 -elicted autophagy and apoptosis, as well as demonstrate the efficiency of metformin in prevention and therapy, following environmental carcinogen exposure.

Nickel induces autophagy via HK2 and LCN2 induction in lung cells
To assess the effects of nickel on cell fate, we evaluated autophagy induction and the generation of relevant proteins in the presence of nickel. Autophagy plays an essential role in lung oncogenesis. At the beginning of autophagy, the cytosolic form of LC3B (LC3B-I, 18 kDa) is converted to the phagophore and autophagosome bound form of LC3B (LC3B-II, 16 kDa). Treatment with various concentrations of nickel for various time periods resulted in cell autophagy in BEAS-2B cells. On western blot, 0.25 mM concentration of nickel significantly induced LC3B-II/ LC3B-I ratio after 48 h in BEAS-2B cells ( Figure 1A and 1B). In a previous study, nickel accumulation increased cellular glycolytic activity, which is the foremost alteration of energy metabolism in tumorigenesis (the Warburg effect) [25,26]. In the present study, NiCl 2 affected multiple genes in the glycolysis pathway on Agilent SurePrint G3 Human V2 GE 8×60K microarray analysis of KEGG pathway. In particular, treatment with NiCl 2 induced a 38-fold increase in HK2 mRNA level. (Supplementary Materials,  Supplementary Table 1 and Supplementary Figure 1). Nickel stimulated HK2 and inflammation protein LCN2 expressions in BEAS-2B cells, as demonstrated on western blot ( Figure 1A and 1B). To assess the mRNA levels of HK2 in NiCl 2 -treated cells, BEAS-2B cells and WI-38 fibroblasts were treated with various concentrations of NiCl 2 for 48 h and analyzed on RT-PCR and Q-PCR ( Figure  1C). To confirm autophagic flux in NiCl 2 -treated cells, AVO development was detected by staining of late autophagic vacuoles with acridine orange dye. As shown in Figure  1D upper, nickel prompted AVO formation in BEAS-2B cells. To calculate the AVO fractional volume after NiCl 2 treatment, flow cytometric analysis was performed. The data indicated that NiCl 2 stimulates AVO development in a dose-dependent manner in BEAS-2B cells ( Figure 1D lower). Secreted LCN2 levels increased following 0.25 mM NiCl 2 treatment in BEAS-2B cells ( Figure 1E). These results implied that HK2 and LCN2 are activated in the presence of NiCl 2 .

Metformin represses autophagy and apoptosis by inhibition of NiCl 2 -induced HK2 levels and activity
HK catalyzes the first step of glycolysis, phosphorylating glucose to glucose-6-phosphate. HK2 plays an important role not only in glycolysis, but also in cell survival. We investigated whether HK2 is involved in NiCl 2 -induced cell fate. NiCl 2 increased HK2 mRNA levels and metformin diminished NiCl 2 -induced HK2 expression in a dose dependent manner (Figure 2A upper). The same results were obtained on real time RT-PCR for the detection of mRNA expression of HK2 in BEAS-2B cells (Figure 2A lower). As shown in Figure  2B, we performed western blotting to examine whether metformin diminishes up-regulation of the protein levels of HK2 in the presence of NiCl 2 . To quantify the HK2 protein expression affected by NiCl 2 and metformin on western blotting, we repeated the same experiment three times. As 2-DG is an analog of glucose, it has been used as an HK inhibitor, as it competes with glucose. Remarkably, 2-DG treatment also decreased NiCl 2induced HK2 expression ( Figure 2C). As shown in Figure 2D, metformin and 2-DG significantly decreased NiCl 2 -induced HK activity. Moreover, the decrease in HK2 expression was based on two different short hairpin RNAs (shRNAs). The responses of BEAS-2B shGFP cells were similar to those of parental BEAS-2B cells after NiCl 2 and metformin treatment. In BEAS-2B shHK2 cells, NiCl 2 -elicited LC3B-II and cleaved caspase-7 expressions were significantly diminished. In addition, there was strong correlation between the expression of HK2 and LCN2 level. Moreover, BEAS-2B shHK2 cells significantly decreased following NiCl 2 treatment ( Figure  2E). To confirm the effect of HK2 silencing on AVO fraction volume after NiCl 2 and metformin treatment, flow cytometry was performed (20.6% versus 7.1% and 10.4%) ( Figure 2F). These data indicated that HK2 is involved in the induction of autophagy in the presence of NiCl 2 . It is well known that the generation of reactive oxygen species (ROS) contributes to nickel-triggered carcinogenesis, including EMT promotion and the cause of DNA damage [8,27]. To determine whether metformin can suppress NiCl 2 -induced ROS accumulation, cells were treated with 2′, 7′ -dichlorodihydrofluorescein diacetate (H 2 DCFDA) and analyzed by flow cytometry. Results revealed that metformin decrease ROS generation in the presence of nickel (10.42% versus 5.58%). N-acetyl-cysteine (NAC, 1 mM), the ROS scavenger, was used to confirm the reversion of NiCl 2 -induced ROS ( Figure 2G).

Endogenous LCN2, but not exogenous LCN2, triggers NiCl 2 -mediated autophagy in bronchial epithelial cells
LCN2, also known as neutrophil gelatinaseassociated lipocalin (NGAL), is required for tumor progression and metastasis. It is often implicated in the responses to hypoxia and apoptosis induction [28]. However, the correlation between LCN2 and autophagy in the presence of NiCl 2 remains unclear. Actually, a causal link between LCN2 and HK2 levels and autophagy levels in bronchial epithelial cells has not been reported, which prompted us to clarify whether LCN2 is involved in NiCl 2 -elicited autophagy. To assess the effect of metformin on NiCl 2 -induced LCN2 expression, BEAS-2B cells were cultured in the presence of NiCl 2 with or without metformin for 48 h, and RT-PCR and qPCR were performed to detect the mRNA expressions of LCN2. As shown in Figure 3A, metformin significantly decreased NiCl 2 -induced LCN2. In addition, treatment with metformin or 2-DG reduced NiCl 2 -elicited LCN2 protein levels. The same results were obtained on ELISA for secretion of LCN2 in a BEAS-2B cell culture supernatant following NiCl 2 , metformin or 2-DG treatment ( Figure 3B, 3C). Particularly, the protein level of LCN2 was downregulated approximately 40% in all cell lysates following 2-DG treatment. However, in metformin-treated cells, NiCl 2 -induced secretion of LCN2 was blocked ( Figure  3B, 3C). These results revealed that NiCl 2 -mediated LCN2 is repressed by metformin at translational and transcriptional levels.
We further examined whether LCN2 independently prompts autophagy and attempted to clarify the relationship between LCN2 and HK2 in lung epithelial cells. Western blot analysis was performed to observe whether LCN2 treatment induces the processing of LC3B-I to LC3B-II, and the protein expression of HK2. The results demonstrated that HK2 and LC3B-II/I do not accumulate following LCN2 treatment (10 ng/mL) for up to 48 h ( Figure 3D). We utilized acridine orange staining to confirm this result. Concentrations of LCN2, up to 40 ng/mL, failed to alter the increase in AVO accumulation ( Figure 3E). Furthermore, the protein levels of HK2 and LC3B-II/I ratios were determined after attenuation of LCN2 activity by small hairpin RNA (shRNA). The data revealed that blockade of endogenous LCN2, but not exogenous LCN2, represses autophagy in BEAS-2B cells ( Figure 3F). To calculate the AVO fractional volume, we performed flow cytometric analysis 48 h after LCN2 silencing and co-treatment with NiCl 2 and metformin ( Figure 3G). Taken together, the results suggested that LCN2 is involved in NiCl 2 -induced autophagy.

NiCl 2 -mediated autophagy decreases following metformin treatment in human bronchial epithelial cells
To determine the intracellular distribution of ionic nickel in the cells, we utilized Ni 2+ -selective fluorescence dye Newport Green TM DCF, which fluoresces when there is binding with ionic nickel ions. As shown in Figure  4A upper, in a dose-dependent experiment, BEAS-2B cells were exposed to various concentrations of NiCl 2 for 48 h. The result was an incremental increase in green fluorescence in NiCl 2 -treated cells. In a previous study, cellular responses of metformin were associated with this drug's metal-binding properties, especially binding with copper. However, the association with nickel remained unclear [29]. To further investigate whether metformin in BEAS-2B cell culture supernatant as revealed on ELISA. The data are expressed as mean±SD. * p<0.05, *** p<0.001 on two-tailed t test compared with control. www.impactjournals.com/oncotarget sequesters intracellular nickel, BEAS-2B cells were co-treated with 0.25 mM NiCl 2 and 5 mM metformin for 48 h, then stained with Newport Green TM DCF. As shown in Figure 4A lower, treatment with metformin did not affect NiCl 2 -activated fluorescence, indicating that metformin does not contribute to blockage of nickel ions into cells. To evaluate the effect of metformin on NiCl 2elicited autophagy, we utilized pEGFP-LC3 transient transfection to visualize aggregation of expression of LC3B. After treatment with 0.25 mM NiCl 2 , GFP-LC3 was redistributed from a ubiquitous, diffuse pattern toward autophagosomes, observed as cytoplasmic dots in BEAS-2B cells. Remarkably, treatment with 5 mM metformin decreased the formation of GFP-LC3 puncta ( Figure 4B). We then examined the effect of metformin on NiCl 2 -induced AVO development. Treatment with metformin blunted AVO formation in a dose-dependent manner in NiCl 2 -treated cells. Flow cytometric analysis was performed to quantify the AVO fractional volume (18.35% versus 6.98%) ( Figure 4C). mTOR, the Akt and AMPK downstream effector, plays a critical role in cell proliferation, growth and survival. Activated mTOR promotes protein translation by phosphorylating its substrates, including p70 ribosomal protein S6 kinase (p70S6K) [30]. To identify the effects of metformin on NiCl 2 -induced autophagy-related pathway and genes, NiCl 2 and metformin were administered for 48 h, followed by analysis of protein expressions on western blotting. As shown in Figure 4D and Figure 4E, Akt-Ser473 phosphorylated level was downregulated, but did not affect phosphorylated p70S6K-Thr389, AMPK-Thr172, beclin-1, ATG12-ATG5 and ATG3 levels after 0.25 mM NiCl 2 treatment for 48 h. Beclin-1, ATG12-ATG5 and ATG3 are members of the ATG family involved in the formation of autophagosomes [31]. Metformin treatment revived the Akt-Ser473 phosphorylated expression, but reduced NiCl 2 -mediated autophagy, perhaps through Aktdependent but mTOR-independent pathway. In addition, metformin is known to activate AMPK. To determine the relationship between HK2 and p-AMPK on nickel-induced autophagy, shGFP and shHK2 BEAS-2B cells treated with NiCl 2 and metformin were performed. As shown in Figure  4F, there were no significant changes in the protein levels of p-AMPK and p-P70S6K after knockdown of HK2.

NiCl 2 -elicted autophagy contributes to activation of apoptosis
It is well documented that nickel exposure induces apoptosis via ROS accumulation and involvement of mitochondria, ER-stress, Fas, and c-Myc [32]. To investigate the effects of metformin on NiCl 2 -mediated apoptosis and to clarify the cellular sources of HK2 in autophagy and apoptosis in NiCl 2 -treated cells, BEAS-2B cells were treated with 1, 2.5 or 5 mM metformin, with or without 0.25 mM NiCl 2, for 48 h and analyzed by flow cytometry. As shown in Figure 5A, NiCl 2 -mediated apoptosis was recovered by metformin (13.22% versus 3.17%). In addition, we assessed the protein markers of apoptosis and autophagy in the presence of NiCl 2, with or without metformin or 2-DG, on western blotting. Metformin or 2-DG treatment simultaneously abolished the increases in LC3-I/LC3-II ratio and cleaved caspase-7 expression in NiCl 2 -treated BEAS-2B cells ( Figure 5B). To further determine whether NiCl 2 -mediated autophagy stimulates apoptosis, the autophagy inhibitor chloroquine (CQ) was used to suppress late phase autophagy. As shown in Figure 5C, treatment with NiCl 2 and metformin, as well as with 10 μm CQ, blocked the endogenous LC3-II turnover and resulted in increased NiCl 2 -elicted autophagy. CQ treatment also inhibited NiCl 2 -mediated cleavage of poly ADP-ribose polymerase and cleavage of caspase-7, which served as apoptotic markers. Particularly, metformin prevented the accumulation of LC3-II and apoptotic proteins, with or without CQ treatment. To further confirm the correlation between autophagy and apoptosis in cells exposed to NiCl 2 , we observed atg5 knockout mouse embryonic fibroblast cells (Atg5 -/-MEF cells), as well as LC3 knocked-down BEAS-2B cells. In atg5 -/and atg5 wild-type MEF cells, NiCl 2 treatment slightly increased the protein expression of HK2. Similar to NiCl 2 -treated BEAS-2B cells, in atg5 WT MEFs there was significant induction of LC3 I to II conversion, as well as cleavage of PARP and caspase 3 expression. However, there was failure to prompt atg5 -/-MEFs ( Figure 5D). As shown in Figure 5E, the specific shRNA targeting LC3 was transfected into BEAS-2B cells with knockdown of the expression of LC3. In comparison with BEAS-2B shGFP cells, NiCl 2 -induced cleavage of caspase 7 was blunted in BEAS-2B shLC3 cells. Our results demonstrated that NiCl 2 -induced autophagy induces apoptosis.

HK2 is the crucial regulator in lung cancer progression
We assessed HK2 and LCN2 expressions in The Cancer Genome Atlas (TCGA) Data Portal from Broad GDAC Firehose and performed immunohistochemical staining to detect the expressions of HK2 and LC3B in 72 human lung cancer specimens to determine whether HK2, LCN2 and LC3B are involved in lung cancer progression. The representative IHC results are shown in Figure 6A. The presence or absence of HK2 and LC3B protein expressions was associated with tumor stage, T status and metastasis (Supplementary Table 2). Both HK2 and LCN2 expressions significantly increased in cancer tissues when compared with normal tissues in lung squamous cell carcinoma ( Figure 6B). Furthermore, we examined the expressions of HK2 and LCN2 in LUSC and LUAD tissues and their corresponding noncancerous tissues using the TCGA Data Portal ( Figure 6C). The results revealed that HK2 and LCN2 are associated with tumor progression, especially in LUSC tissue.

DISCUSSION
It is well documented that metformin alleviates autophagy and apoptosis via HK2 and LCN2, following exposure to nickel, in bronchial epithelial cells. The molecular basis of nickel carcinogenicity has proven complex, as many chronic inflammation and stress response pathways are activated in nickel-specific toxicology profiles. There is much evidence that chronic inflammation contributes to the onset and progression www.impactjournals.com/oncotarget of cancer [33,34]. In recent studies, Toll-like receptor 4 (TLR4) has been identified as the critical mediator of the innate immune response to nickel that triggers NF-κB signaling and pro-inflammatory gene activation [35]. Although nickel compounds have low mutagenic capabilities, previous studies have found that nickel accumulation in lung tissues contributes to incremental levels of EGFR and P53 mutations, which can reduce DNA repair activity and promote tumor invasion, leading to lung carcinogenesis [36][37][38]. In addition, accumulating evidence has emphasized the importance of nickel in modulating the epigenetic landscape that includes chromatin structural modifications, DNA methylation and histone modifications [39,40]. Nickel also induces the upregulation of a specific set of proteins and microRNAs (miRNAs), leading to altered DNA methylation and histone modification landscapes in a variety of cell types [7,39]. From the results of recent studies, HK2 is highly expressed in various cancers, and is regulated by miRNA [41]. MiR-143, an anti-oncomiR, is often downregulated in cancers, such as colon and gastric cancers, as well as B-cell lymphoma [42][43][44]. It targets HK2 mRNA and inhibits HK2 expression [41,45]. Moreover, other miRNAs may be involved in the altered expression of HK2 in tumors, including miR-181b, miR-125b and miR-182 [46][47][48].
Nickel-induced carcinogenesis may involve glycolysis pathway activation. It has been shown that genes related to glucose metabolism and glycolysis are inducible by nickel exposure in an HIF-dependent manner [49]. Actually, HIF has been found to accumulate in various cell lines in the presence of nickel [50]. Increasing numbers of studies have shown that metabolic enzymes directly contribute to carcinogenesis. In comparison with normal tissues, cancer cells prefer to metabolize glucose into lactic acid by glycolysis, which is known as the "Warburg effect", and is accompanied by upregulation of HK2 [9]. Previous studies have demonstrated that HK2 is highly present in lung and breast cancers, and is required for tumor initiation and maintenance. Tumor progression is impaired following its downregulation [51]. Here, we demonstrated that reduction in NiCl 2 -induced HK2 by metformin inhibits NiCl 2 -mediated autophagy. It is known that metformin suppresses hypoxia-induced HIF-1α accumulation [52]. In this study, we confirmed that metformin decreases nickel-induced HIF-1α expression (data not shown). We also demonstrated the importance of HK upregulation in nickel-induced autophagy through inhibition of the expression and activity of HK2 using competitive HK inhibitor 2-DG and HK2-specific shRNA silencing. HK2 silencing combined with metformin demonstrated that autophagy is not only inhibited by HK2 activation ( Figure 2E).
A previous study demonstrated that HK2 positively regulates protective autophagy via TORC1 inhibition in response to glucose deprivation [14]. However, in contrast with normal tissues, inhibition of HK2 by 2-DG suppresses lung cancer cell growth through induction of cell apoptosis and autophagy [53]. Furthermore, treatment with 2-DG and CQ represses HK2-mediated Warburg effect and ULK1-dependent autophagy activates apoptosis to cause tumor regression [54]. These studies indicated that HK2 is able to regulate different effects of autophagy and is a key mediator and energy precursor. Therefore, we investigated metformin as a new anti-autophagy drug by targeting Ni-accumulated HK2 in lung epithelial cells.
In our previous study, treatment with NiCl 2 stimulated EMT via HIF-1α-dependent pathway and E-cadherin promoter hypermethylation in bronchial epithelial cells [8]. EMT is an important step in the progression of lung cancer toward metastasis and invasion and occurs during the development of epithelial carcinogenesis [55,56]. Actually, the correlation between EMT and autophagy has been well studied over the past decade. It has been observed that autophagy contributes to cancer invasion through EMT activation during starvation or hypoxia [57][58][59].
A recent report showed that HK2 overexpression is associated with the hypomethylation status of CpG island region -379 to +209 from HK2 promoter in hepatocellular carcinomas [60]. It had been stated that DNA hypomethylation of proto-oncogene contribute to nickel-induced malignant transformation [61]. We could not rule out the possibility that nickel will regulate the epigenetic alteration of HK2 promoter hypomethylation. In addition, metformin may obstruct HIF-1α binding on HK2 promoter resulting in HK2 decrease. Overall, these results evidence that the potential capacity of nickel alter the epigenetic regulation and the transition to the status of higher aggressiveness and thus progression of carcinogenesis.
It is worth noting that autophagy induced by TLR4 signaling promotes TLR-triggered cytokine production, which accelerates migration and invasion of lung cancer cells [62]. Studies have suggested that LCN2 upregulated by LPS is influenced via TLR4 signaling pathway [63,64]. In this study, we demonstrated that NiCl 2 induces autophagy induction via increment levels of HK2 and LCN2. As shown in Figure 2F and Figure 3F, we used acridine orange stain to verify the role of HK2 and LCN2 in nickel-induced autophagy after HK2 and LCN2 gene silencing. The results demonstrated that NiCl 2-elicited autophagy is via HK2-LCN2 pathway. Although we suggested that LCN2 is involved in NiCl 2 -induced autophagy pathway, we could not clarify whether nickel upregulates LCN2 expression via TLR4 signaling or if nickel-induced autophagy and EMT activation are facilitated by TLR4 at the source. We did find that metformin restores the protein expression of E-cadherin (data not shown). In future studies, we will explore the precise mechanism of metformin in diminishing the various effects of nickel.
Nickel induced malignant transformation through SQSTM1/P62 and inflammatory TNF upregulation in Beas-2B cells [65]. Son et al. found that nuclear factor erythroid 2-related factor 2 (Nrf2) plays an important role in nickel-induced autophagy [66]. In the present study, we found that nickel induce autophagy via HK2 and LCN2. Autophagy is triggered by the stress of metabolism and inflammation. Our results suggest that autophagy is one of cancer-promoting reasons under nickel exposure. In fact, autophagy has been shown to play a protective role against apoptosis in malignant transformation. Whether autophagy promotes cell survival or cell death depends on the levels of stress. When stress severity or duration increases, cell death may result. In brief, autophagy affects cellular homeostasis [67,68]. From our data, nickel simultaneously induces autophagy and apoptosis. To further demonstrate the association of autophagy with apoptosis in the presence of nickel, endogenous LC3 was knocked down by shRNA treatment, and WT or Atg5 -/-MEF cells were treated with CQ ( Figure 5C-5E). We observed that the activation time point of nickelinduced autophagy is earlier than that of apoptosis (data not shown). We used short-term exposure and high concentrations of nickel to clarify the role of nickelinduced autophagy. In the presence of excessive nickel, there is an imbalance in autophagy, which promotes apoptosis.
There are some limitations in the study. Nickel contents were hardly determined in tissue array. We could not analyze the samples for the relationship of nickel exposure and HK2 expression. However, we investigate the tumorigenesis of HK2 and LCN2 on TCGA database ( Figure 6B, 6C). Both HK2 and LCN2 serve as biomarkers in lung cancer progression. We also observed the expression of HK2, LCN2 and autophagy-related genes in lung cancer cell lines, including CL1-0, CL1-5, TL-6 and H1975 with or without 0.25 mM NiCl 2 (Supplementary Figure 2A, 2B). Results revealed that most of cancer cells elevate the expression of HK2 and LC3B in the presence of nickel. Equivalently with BEAS-2B cells, autophagyrelated genes Atg5 and Beclin-1 except LC3B were decreased after NiCl 2 treatment.
In conclusion, the results of this study provide evidence that metformin alleviates NiCl 2 -stimulated autophagy via the inhibition of HK2 and LCN2 expressions (Figure 7). Accumulation of nickel triggers metabolic changes and inflammatory environment contributes to lung cancer development. The results of this study also demonstrated the preventive effects of metformin against cumulative damage caused by environmental carcinogens.

Human LCN2 enzyme-linked immunosorbent assay (ELISA)
ELISA was performed using Human LCN2/NGAL DuoSet ELISA Kit (R&D Systems, DY1757) according to the manufacturer's instructions. The absorbance at 450 nm was measured using a microplate reader.

Detection and quantification of acidic vesicular organelles with acridine orange
Autophagy is the process of packaging cytoplasmic proteins into the lytic component and characterized by the formation of acidic vesicular organelles (AVOs). After treatment for 48 h, cells were washed in PBS and stained with acridine orange (Sigma, A6014) (1 μg/ml) in serumfree LHC-9 medium for a period of 15 min, then washed twice with PBS and suspended in LHC-9. To observe the formation of AVOs, the cells were detected under a red filter fluorescence microscope and quantified using flow cytometry.

HK activity assay
HK was assayed according to the manufacturer's instructions using HK Colorimetric Assay Kit (Biovision, K789-100). To assay total HK activity after treatment with NiCl 2 and metformin, cell lysates were homogenized with ice cold HK Assay Buffer and the supernatant was collected with reaction mixture. This was followed by incubation for 5 min at room temperature and measurement at excitation wavelength of 450 nm for 30 min. Specific activity was determined using NADH standard. All experiments were repeated at least three times. www.impactjournals.com/oncotarget VZV-G pseudotyped lentivirus-shRNA system RNAi reagents were obtained from the National RNAi Core Facility located at the Institute of Molecular Biology/Genomic Research Center, Academia Sinica. Individual clones were identified by their unique TRC number: shGFP TRCN0000072178 (responding sequence: CAA CAG CCA CAA CGT CTA TAT) and shLuc TRCN0000072246 (responding sequence: CAA ATC ACA GAA TCG TCG TAT) for vector control; shHK2 (27)

Detection of intracellular nickel
Intracellular mobilized ionic nickel was detected after NiCl 2 and metformin treatment using green fluorescent Newport Green TM DCF diacetate indicator dye (Invitrogen, N7991). The treated cells were washed twice with HBSS and incubated with 1 μM Newport Green TM DCF diacetate in LHC-9 for 30 minutes. Then, they were washed twice with HBSS after recovery in LHC-9 containing Newport Green TM DCF diacetate. Green fluorescence was visualized under an Olympus CK40 fluorescence microscope.

pEGFP-LC3 plasmid transfection
pEGFP-LC3 expression vector was purchased from Addgene (#21073). The pEGFP-LC3 plasmid is a pEGFP-C1 plasmid inserted into microtubuleassociated protein 1 light chain 3 (LC3) cDNA at the C-terminus and green fluorescent protein (GFP) at the N-terminus [69]. pEGFP-LC3 fusion protein was used to visualize the autophagosomes in cells. Transfection was performed on 24-well plates with coverslips, with 1 μg plasmid in each well and jetPEI transfection reagent (Polyplus-transfection, 101-10). This was followed by incubation overnight. The medium was removed and fresh medium containing NiCl 2 and metformin was added to the wells for 48 h. After exposure, the cells were washed twice with PBS and fixed in 3.7% paraformaldehyde-PBS for 10 min at room temperature. Observation of GFP-LC3 puncta in cells was carried out under confocal microscope (ZEISS LSM510 META).

Detection of intracellular reactive oxygen species (ROS)
Cellular ROS was detected using the fluorescence probe 2′, 7′-dichlorodihydrofluorescein diacetate (H 2 DCFDA) (Invitrogen, D399). After nickel and metformin treatment for 48 h, BEAS-2B cells were stained with 20 μM H 2 DCFDA at 37 °C for 30 min in the dark. Then washed twice with PBS and harvested cell in PBS contained with 5% FBS. The fluorescence intensity was analyzed using flow cytometry.

Expression analysis of the cancer genome atlas lung squamous cell carcinoma and lung adenocarcinoma data
Gene expression data were obtained from The Cancer Genome Atlas (TCGA) lung squamous cell carcinoma (LUSC) and lung adenocarcinoma (LUAD) datasets (https://tcga-data.nci.nih.gov/tcga). The datasets contain data from 463 LUSC samples with 50 adjacent normal tissue samples and 488 LUAD samples with 57 adjacent normal tissue samples, respectively.

Immunohistochemistry
Antibodies against HK2 (2867) and LC3B (3868) were obtained from Cell Signaling Technology. Negative controls were used, leaving out the primary antibody. Immunohistochemical methods were carried out using conventional streptavidin peroxidase method according to the manufacturer's (Dako) LSAB Kit (K675) procedure. Slides were visualized using 3,3'-diaminobenzidine tetrahydrochloride as a substrate. The control slide (Dako, T1076) and semi-quantitative H scores of HK2 immunoreactivity were determined by multiplying the proportional scores of stained cells by their immunoreactivity intensity. All immunohistochemical staining cases were examined by two pathologists (Pei-Ru Wu and Kun-Tu Yeh, Department of Pathology, Changhua Christian Hospital, Changhua, Taiwan), and a final agreement was obtained for each score at a discussion microscope.

Author contributions
Yu-Ting Kang and Wen-Cheng Hsu: performed experiments, analyzed data and literature search.
Chih-Hsien Wu, I-Lun Hsin and Pei-Ru Wu: analyzed data and contributed to writing of the manuscript.
Kun-Tu Yeh and Jiunn-Liang Ko: designed the study and supervised the manuscript preparation.
All authors were involved in preparing the paper and had final approval of the submitted and published versions.