Low levels of AMPK promote epithelial‐mesenchymal transition in lung cancer primarily through HDAC4‐ and HDAC5‐mediated metabolic reprogramming

Abstract AMP‐activated protein kinase (AMPK) serves as a “supermetabolic regulator” that helps maintain cellular energy homeostasis. However, the role of AMPK in glucose metabolism reprogramming in lung cancer remains unclear. Here, our study shows that low AMPK expression correlates with metastasis and clinicopathologic parameters of non–small‐cell lung cancer. Low AMPK significantly enhances the Warburg effect in HBE and A549 cells, which in turn induces the expression of mesenchymal markers and enhances their invasion and migration. At the mechanistic level, low AMPK up‐regulates HK2 expression and glycolysis levels through HDAC4 and HDAC5. Collectively, our findings demonstrate that low AMPK‐induced metabolism can promote epithelial‐mesenchymal transition progression in normal bronchial epithelial cells and lung cancer cells, and increase the risk for tumour metastasis.


| INTRODUC TI ON
Lung cancer has become a leading cause of cancer-related deaths in China and worldwide. Deaths of most lung cancer patients are metastasis-related. 1 In recent years, with the advancement of radical surgical techniques, the advent of new molecular targeted drugs and the development of neoadjuvant therapy, prognosis of lung cancer patients has significantly improved. 2 However, 5-year overall survival in patients with lung cancer is still unsatisfactory. Therefore, seeking new ways to inhibit lung cancer metastasis is a common objective of thoracic experts.
In addition to tumour cells' infinite proliferative capacity, loss of contact inhibition and migration ability, metabolic rearrangement is yet another important feature of malignant tumours. 3,4 To facilitate proliferation, invasion and escape from the immune system, tumour cells actively choose glycolysis as the main way to obtain energy, even in an oxygen-rich environment, namely the Warburg effect.
Due to the heterogeneity of cancer, there are significant differences in the metabolic profiles of different types of tumours, and tumours at different stages of development, further complicating things. 5,6 Epithelial-mesenchymal transition (EMT) is a process in which the expression of epithelial markers, such as E-cadherin decreases and the expression of interstitial markers (eg, N-cadherin and Fibronectin) increases. [7][8][9] For non-small-cell lung cancer (NSCLC), it is ubiquitous for tumour cells to acquire increased mobility, invasiveness and apoptosis resistance through pathological EMT. 10,11 In some sense, EMT is also an adaptive behaviour of tumour cells, exchanging cellular demands for rapid proliferation for survival and metastasis. It is therefore plausible to assume that the EMT process and metabolic reprogramming of tumour cells are closely related.
Studies have shown increased glucose uptake and metabolic enzyme expression in EMT-transformed breast, lung and pancreatic cancer cells. However, the metabolic control points, the specific relationship between metabolic reprogramming and EMT in NSCLC cells, and related mechanisms still remain poorly defined. AMP-activated protein kinase (AMPK) serves as a "supermetabolic regulator" that helps maintain cellular energy homeostasis. 12 It is a heterotrimeric protein, consisting of a catalytic α-subunit (isoforms α1 or α2), a regulatory β-subunit (isoforms β1 or β2) and a nucleotide binding γ-subunit (isoforms γ1, γ2 or γ3). The α-subunit has a conventional serine/threonine kinase domain at the N-terminus, with a conserved threonine residue. 13 AMPK has been associated with several physiological processes including cell division, endothelial cell migration and maintenance of epithelial cells polarity. From a metabolic perspective, AMPK promotes ATP conservation, under conditions of metabolic stress. It activates catabolic metabolism pathways, such as autophagy, and inhibits anabolic processes. 14,15 A recent report by Jones et al 16 showed that AMPK is a negative regulator of the Warburg effect and that it suppresses tumour growth. Despite evidence linking AMPK with tumour suppressor functions, the role of AMPK in tumorigenesis and tumour metabolism is still not known.
In this study, we specifically sought to address the following aspects: (1) what is the involvement of AMPK in tumorigenesis and progression of NSCLC; and (2) what are the mechanisms associated with AMPK's effects on tumour metabolism and tumour metastasis.

| MATERIAL S AND ME THODS
All experiments were performed in adherence with the National Institutes of Health (NIH Publication, 8th Edition, 2011) guidelines on the use of laboratory animals. The animal care and experimental protocols were approved by the Suzhou University Committee on Animal Care.
A series of human NSCLC specimens were obtained from the pathology department of affiliated hospital of Suzhou University.
The patients' clinical information includes age, sex, tumour differentiation, lymph node metastasis and TNM stage. All specimens were collected under the guidance of the HIPAA protocol and supervised by the ethics committee of the hospital.

| Construction of tissues microarrays and immunohistochemistry
Tissues microarrays (TMA) were constructed by the Department of Pathology, the First Affiliated Hospital. For immunohistochemistry (IHC), after antigen retrieval using EDTA, the specimens were blocked with goat serum for 20 minutes before applying the primary antibody. Specimens were incubated with anti-AMPKα (Abcam, cat #ab32047; 1:200) for 12 hours in 4°C. Next, the sections were washed twice and subsequently incubated with HRPpolymer-conjugated secondary antibody (Zhong Shan, China) at room temperature. Finally, the microarrays were then stained with 3, 3-diaminobenzidine solution and haematoxylin. The slides were photographed with an inverted microscope (Olympus).

| Animal feeding and treatment
BALB/c nude mice (at 6-8 weeks of age) were purchased from Beijing HFK Bio-technology. Mice were housed in a controlled environment (humidity, 50%-60%). A total of five mice were housed per cage and were maintained at room temperature under a 12-h light/dark cycle; Mice were provided free access to food and water.
Plasmids and the RNAi vector were successfully constructed and then packaged in 293T cells. The concentrated titre of virus suspension was 4 × 10 12 Tu/L.

| Establishment of stable cell lines
0.8 × 10 6 A549 cells were seeded into 60-mm plastic dishes. After the cells reached about 30%-35% confluence, lentiviruses containing AMPK siRNA (sgAMPK) or HK2 siRNA were infected following the manufacturer's protocol at the desired multiplicity of infection (MOI = 10). After 8 hours, infection medium was removed and fresh medium was added. After an additional 64 hours, GFP co-expression on the construct was used to determine efficiency of viral transduction.

| Cellular proliferation, invasion and migration assays
In brief, CCK-8 assay was applied to measure the cell proliferation according to the Cell Counting Kit-8 manufacturer's protocol (Meilunbio, catalog #MA0218). For migration and invasion assays, the transwell filter inserts with a pore size of 8 μm were coated without or with matrigel (1:10 dilution), respectively. 5 × 10 4 cells (for migration) and 1 × 10 5 cells (for invasion) were, respectively, seeded in serum-free medium in the upper chamber. After 24 h incubation at 37°C, cells in the upper chamber were carefully removed with a cotton swab and the cells that had traversed the membrane were fixed in methanol, stained with Crystal violet (0.04% in water; 100 μL) and counted the permeating cells under the inverted microscope and photographed.

| Apoptosis
Apoptosis assay was carried out using the Annexin VFITC/PI apoptosis detection kit (KeyGen Biotech) according to the manufacturer's protocol. Briefly, 0.5 mL binding buffer was added into 1 × 10 5 A549 and HBE cells. Sequentially, the cells were stained with PI at room temperature for 15 min and then analysed by flow cytometry (BD, FACSCantoTM II).
Western blots were quantified using ImageJ software. Protein levels were calculated from the ratio of corresponding protein/β-actin.

| RNA extraction and RT-PCR analysis
TRIzol reagent was used to extract RNA from A549 and HBE cells. The cDNA was generated with random primers using the Reverse Transcription System (Promega). GAPDH was used for normalization of qRT-PCR data.
Primer sequences used in this study were listed in Table S2.

| Extracellular flux measurements
Metabolic analyses were done with the Seahorse XFe96 Analyzer (Seahorse Bioscience), which measures the extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) of live cells.
Glycolytic rates were measured with the Seahorse XF glycolytic rate assay (S7805A, Seahorse Agilent).

| High performance liquid chromatography
Cells were quickly extracted with pre-cooled 80% methanol and assayed using a high performance liquid chromatography (HPLC) system. ATP levels were calculated by dividing the peak area of samples by standards.

| Statistical analysis
Numerical data are expressed as the mean ± SEM. Two independent sample data sets were tested using two-tailed Student's t test.
Multiple group comparisons were evaluated by one-way ANOVA followed by least significant difference t test for post hoc analysis.
Chi-square or Fisher's exact tests were used to compare categorical variables. Analyses were performed using SPSS software (SPSS, Inc.). P < .05 was considered as significant difference.

| Low AMPK expression correlates with clinicopathologic parameters of NSCLC
To investigate the expression of AMPK in lung cancer, we constructed tissue microarray (TMA) of 192 human NSCLC specimens, followed by immunohistochemical (IHC) analysis ( Figure 1A).
The correlation between AMPK level and the clinicopathologic  (Table S3). Compared with histology grade 1, AMPK level in histology grade 3 was significantly lower (P = .022, χ 2 test; Figure 1B). The results also indicated that the low expression of AMPK was positively correlated with lymph node metastasis (P = .016, χ 2 test, Figure 1D) and tumour T stage (P = .026, χ 2 test, Figure 1C), but not with epidermal growth factor receptor (EGFR) mutation rate (P > .05, χ 2 test, Figure 1E). Interestingly, the models harvested after 10 weeks showed significantly higher probability of chest wall metastasis in the siAMPK-LV-treated group (P = .02, χ 2 test), suggesting that low AMPK expression might be promoting lung cancer metastasis ( Figure 2I).

| Low AMPK induces EMT in HBE cells
Our clinical data and result of the tumour xenograft study suggest that low AMPK expression is associated with lung cancer metastasis. We subsequently performed transwell assay to determine the effect of AMPK on A549 and HBE cells' migration and invasion. Overexpression and knockdown treatments of AMPK were then applied to A549 cells. The results showed that siAMPK-LV treatment significantly increased migration and invasion of HBE  Figure S1). However, in HBEs cells, overexpression of AMPK had no effect on proliferation activity and apoptosis ( Figure S2).

| Low AMPK induces cell EMT by promoting glycolysis in HBE cells
Considering that AMPK is a key factor, regulating cell energy balance, we tried to determine the relationship between siAMPK-LV and ATP production in HBE cells. We found that treatment of

| AMPK regulation of hexokinase 2 expression and of metabolism requires the involvement of histone deacetylases 4 and 5
The histone deacetylases (HDACs) are known to be involved in the transcription of various signals in the nucleus, including those related to metabolic pathways. To determine whether HK2 gene expression was regulated by HDACs, A549 cells were exposed to a single 60-

| D ISCUSS I ON
For patients with lung cancer, inhibiting the metastasis of tumour cells means extending their survival time and improving their quality of life. It is recognized that the metabolic characteristics of tumours present significant traits that promote proliferation, metastasis and immune evasion. The major finding in this study is that AMPK is an important regulator of lung cancer cells' metabolism and is closely related to the histological grade and malignancy of these tumours.
In order to meet the needs of growth, survival and metastasis, cancer cells employ distinct metabolic pathway to acquire prodigious anabolic materials and products. 17 Termed the Warburg effect, cancer cells deploy high glycolysis and glucose catabolism to produce ATP and lactic acid. 18,19 In addition, the formation of a tumour microenvironment that is resistant to the immune system defences, also depend on this particular metabolic approach. 20,21 Therefore, it is widely accepted that targeting tumour metabolism is a very promising treatment approach for cancer. Disturbingly, many metabolic pathways are shared between tumours and normal cells.
Hence, disrupting the metabolism of tumour cells will inevitably be damaging normal cells. Such is the consequence of treatment with DON (6-Diazo-5-oxo-L-nurleucine), a glutamine antagonist, which exhibits extraordinary anti-tumour effects and unacceptable side effects. 22 According to reports, AMPK serves as a metabolic tumour inhibitor, which can reduce the risk and/ or mortality of certain types of cancer, especially breast cancer, pancreatic cancer and prostate cancer. In the context of tumour development, cells of epithelial origin would transform into a mesenchymal-like phenotype to obtain greater mobility, invasiveness and anti-apoptosis. In this work, we present a comprehensive set of data, suggesting remarkable roles for AMPK in EMT, metabolic regulation and progression of lung cancer. Specifically, we first show that the regulator, AMPK, which is responsible for maintaining cellular energy homeostasis, has lower expression in A549 cells than in normal bronchial epithelial HBE cells. Second, siAMPK treatment up-regulates the expression of various glycolysis-limiting enzymes, including HK2, and subsequently increases the rate of glycolysis and lactic acid production in HBE cells. As expected, artificially up-regulating the expression of AMPK significantly inhibited the level of glycolysis in A549 cells. It should be emphasized that knockdown of AMPK in F I G U R E 4 Low AMPK induces cell EMT by promoting glycolysis in HBE cells. A, High-performance liquid chromatography (HPLC) analysis for ATP in HBE cells, **P < .01 vs the vector group; NS, P > .05 vs the control group, n = 3. B, Relative glycolysis rates in HBE cells, as judged by Seahorse analyses, **P < .01 vs the vector group; NS, P > .05 vs the control group, n = 3. C, D, Metabolic analyses were done with the Seahorse XFe96 Analyzer, which measures the ECAR and OCR of live HBE cells, ***P < .001 vs the vector group; NS, P > .05 vs the control group, n = 3. E, F, Western blot determination of EMT-related markers in response to 2-DG, n = 3. G-L, Relative mRNA expression levels of EMT markers and inducer in HBE cells, *P < .05, **P < .01, ***P < .001, NS, P > .05 vs the indicated group, n = 3 | 7799 FENG Et al.
F I G U R E 5 Low AMPK promotes glycolysis by up-regulating the expression of HK2 in HBE cells. A, B, Relative mRNA and protein expression levels of HK1, HK2, LDHA, G6PI,TPI, MCT-1 and PFK-1 in HBE cells, *P < .05, ***P < .001, vs the Vector group, n = 3. C, D, Relative mRNA and protein expression levels of HK2, AMPK and p-AMPK in HBE cells. E-G, Relative glycolysis rates, migration, invasion in response to siHK2 treatment, *P < .05, **P < .01, ***P < .001 vs the respective vector group in HBE cells, n = 3. H, Western blot determination of EMT-related markers protein expression in response to siHK2 treatment in sgAMPK HBE cells, *P < .05, **P < .01, ***P < .001 vs the vector group, n = 3. I-K, Relative glycolysis rates, migration, invasion in response to siHK2 treatment, *P < .05, **P < .01, ***P < .001 vs the respective vector group in sgAMPK HBE cells, n = 3 F I G U R E 6 AMPK regulates the expression of HK2, and metabolism requires the involvement of HDAC4 and HDAC5. A, B, Relative mRNA and protein expression levels of AMPK and HK2 in response to scriptaid, *P < .05, **P < .01, NS, P > .05 vs the indicated group, n = 3. C, RT-PCR analysis of HDACs at mRNA level in A549 cells with siAMPK and AMPK overexpression treatments, *P < .05, **P < .01 vs the vector group, n = 3. D, Western blot determination of HDAC1, HDAC4 and HDAC5 in A549 cells with siAMPK and AMPK overexpression treatments, n = 3. E, F, Relative mRNA and protein expression levels of HK2 in response to siHDAC4 and siHDAC5, *P < .05, **P < .01 vs the vector group, n = 3. G-I, Relative glycolysis rates, migration, invasion in response to siHDAC4 and siHDAC5 in A549 cells with sgAMPK treatment, *P < .05, **P < .01, ***P < .001 vs the respective vector group, n = 3 A549 cells, further promoted the Warburg effect, allowing these tumour cells gain greater invasiveness. In contrast, overexpression of AMPK led to a reversal of EMT. In addition, it could be that the high level of AMPK in HBE cells, as a result of AMPK overexpression treatment, does not cause significant changes in AMPK at the protein level. This explains, to some extent, why overexpression of AMPK does not affect the proliferative activity and apoptosis of normal bronchial epithelial cells. Third, metabolic reprogramming is a critical step in meeting the demand of tumour cells proliferation, metastasis and immune evasion. Recent studies have shown that EMT can cause changes in a variety of metabolic enzyme activity and their expression levels, to reshape metabolic pathways. 23 Metabolic reprogramming can also be fed back into the EMT mechanism, accompanied by the acquisition of enhanced migration and invasion ability. 24  shows that AMPK regulates GLUT4 transcription in skeletal muscle through HDAC5. There are also reports showing that AMPK-HDAC5 pathway facilitates accumulation of HIF-1a in the nucleus and functional activation of HIF-1 by deacetylating Hsp70 in tumour cells. 28 Here, HDAC4 and HDAC5 were shown to mediate the regulation of AMPK on tumour metabolism. Given the functional characteristics of HDACs, we believe that neither HDAC4 nor HDAC5 directly leads to an increase in HK2 transcription. We suggest that a key intermediate must exist between HDACs and HK2, mediating the regulation of HK2 expression levels and tumour metabolism by HDAC4 and HDAC5. It might be HIF, but more research is needed to support or refute this view.
To summarize, in this study we have demonstrated the remarkable efficacy of AMPK in regulating the metabolism and metastasis of lung cancer cells. At the mechanistic level, low AMPK up-regulates HK2 expression and glycolysis levels through HDAC4 and HDAC5.
Low AMPK-induced metabolism can promote EMT progression in normal bronchial epithelial cells and lung cancer cells, which in turn increases the risk for tumour metastasis. Although more details are still need to elucidate the mechanism, our data uncovered the importance of AMPK in lung cancer progression.

ACK N OWLED G EM ENTS
None.

CO N FLI C T O F I NTE R E S T
The authors report no relationships that could be construed as a conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets used and/or analysed during the current study are available from the corresponding author on reasonable request.