ING5 inhibits aerobic glycolysis of lung cancer cells by promoting TIE1-mediated phosphorylation of pyruvate dehydrogenase kinase 1 at Y163

Abstract

Aerobic glycolysis is critical for tumor growth and metastasis. Previously, we have found that the overexpression of the inhibitor of growth 5 (ING5) inhibits lung cancer aggressiveness and epithelial–mesenchymal transition (EMT). However, whether ING5 regulates lung cancer metabolism reprogramming remains unknown. Here, by quantitative proteomics, we showed that ING5 differentially regulates protein phosphorylation and identified a new site (Y163) of the key glycolytic enzyme PDK1 whose phosphorylation was upregulated 13.847-fold. By clinical study, decreased p-PDK1Y163 was observed in lung cancer tissues and correlated with poor survival. p-PDK1Y163 represents the negative regulatory mechanism of PDK1 by causing PDHA1 dephosphorylation and activation, leading to switching from glycolysis to oxidative phosphorylation, with increasing oxygen consumption and decreasing lactate production. These effects could be impaired by PDK1Y163F mutation, which also impaired the inhibitory effects of ING5 on cancer cell EMT and invasiveness. Mouse xenograft models confirmed the indispensable role of p-PDK1Y163 in ING5-inhibited tumor growth and metastasis. By siRNA screening, ING5-upregulated TIE1 was identified as the upstream tyrosine protein kinase targeting PDK1Y163. TIE1 knockdown induced the dephosphorylation of PDK1Y163 and increased the migration and invasion of lung cancer cells. Collectively, ING5 overexpression—upregulated TIE1 phosphorylates PDK1Y163, which is critical for the inhibition of aerobic glycolysis and invasiveness of lung cancer cells.

Keywords

ING5 / aerobic glycolysis / PDK1 / phosphorylation / lung cancer / TIE1

1 Introduction

Lung cancer is the leading cause of cancer-related death worldwide. Metastasis is the primary reason for the high mortality of lung cancer [1]. Therefore, efficient prevention of cancer invasion and metastasis is necessary to improve the prognosis of patients with lung cancer.

Inhibitor of growth (ING) 5 (ING5) is the new member of the ING candidate tumor suppressor family. It contains a highly conserved C-terminal plant homeodomain (PHD) finger as a binding motif to DNA and proteins. In addition, ING5 associates with HBO1 and MOZ/MORF to form two histone acetyltransferase (HAT) complexes [2]. Therefore, ING5 plays a role in regulating cellular functions, such as proliferation, senescence, DNA damage repair, stem cell self-renewal, and chromatin modification [3–8]. Previously, we demonstrated that ING5 overexpression markedly inhibits the aggressiveness and epithelial–mesenchymal transition (EMT) of lung cancer cells via the inhibition of the epidermal growth factor receptor (EGFR)/phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)/protein kinase B (Akt), interleukin (IL)-6/signal transducer, and activator of transcription 3 (STAT3) and WNT/beta-catenin oncogenic pathways [9–11]. A mechanistic study showed that ING5 overexpression differentially regulates protein acetylation in lung cancer A549 cells, with decreased the acetylation of proteins involved in metabolism [12], indicating the regulatory role of ING5 in cancer metabolic reprogramming.

Metabolic reprogramming is a primary hallmark of cancer cells. It is characterized by increased aerobic glycolysis, which is known as the “Warburg effect,” in which cancer cells prefer to metabolize glucose by glycolysis instead of oxidative phosphorylation (OXPHOS) [13,14], regardless of whether oxygen supply is sufficient [15]. Pyruvate is a metabolic intermediate converted from glucose, which links glycolysis and OXPHOS. The fate of pyruvate decides the metabolic pathway that the cells use to obtain energy [16]. At the junction of glycolysis and OXPHOS Krebs cycle in cellular metabolism, two enzymes compete for pyruvate substrate, including pyruvate dehydrogenase (PDH) and lactate dehydrogenase (LDH) [17]. PDH E1 subunit alpha 1 (PDHA1) is the key component in the PDH multienzyme complex, and it catalyzes the mitochondrial oxidative decarboxylation of pyruvate to acetyl-CoA [18], whereas LDHA converts pyruvate to lactate in the cytosol [19]. In normal cells, pyruvate is carried into the mitochondria when oxygen is available, where it is catalyzed by PDHA1 into acetyl-coenzyme A, which is the rate-limiting step in glucose OXPHOS [20,21]. However, in cancer cells, pyruvate is preferentially converted by LDH-catalyzed fermentation to lactic acid in the cytoplasm, regardless of whether oxygen is sufficient. Thus, PDHA1 serves as a gatekeeper to control the direction of substrate entry into the mitochondrial tricarboxylic acid cycle or cytoplasmic glycolysis [22].

PDHA1 activity could be phosphorylated and inactivated at S293 by pyruvate dehydrogenase kinase 1 (PDK1), leading to the inhibition of pyruvate decarboxylation and OXPHOS and increased pyruvate glycolysis [23]. Therefore, PDK1 functions upstream of PDHA1 as a switching molecule: PDK1 kinase inactivation promotes OXPHOS, whereas PDK1 kinase activation promotes glycolysis [24,25]. PDK1 kinase activity is activated by tyrosine phosphorylation at Y136, Y243, and Y244 [26]. However, the underlying mechanisms mediating the inhibitory regulation of PDK1 kinase remain unknown.

In the current study, the dynamic changes of ING5-regulated phospho-proteomics were observed, and a new tyrosine (Tyr, Y) site Y163 of PDK1 was identified, whose phosphorylation negatively regulates PDK1 kinase activity and is critical for the reversal of the Warburg effect and invasiveness of lung cancer cells by ING5. Increased PDK1 Y163 phosphorylation in clinical lung cancer tissues was correlated with a better prognosis. Furthermore, TIE1 was identified as the upstream kinase for PDK1 Y163, whose level was upregulated by ING5 overexpression. Collectively, a novel mechanism of ING5 was uncovered to reverse the Warburg effect and inhibit lung cancer malignancy.

2 Materials and methods

2.1 Cell culture and reagents

The human lung cancer A549 cell line and human colon cancer HCT116 cell line were purchased from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). ING5 stable overexpression cell lines were established as described previously [9]. PDK1 expression plasmid and PDK1 Y163F (phenylalanine, F) mutation plasmid were constructed by Biowit Biotech (Shenzhen, China). A549 ING5 overexpression cell lines with PDK1 KD, PDK1 WT, and PDK1 Y163F were established by Biowit Biotech. TIE1 knockdown plasmids (shTIE1-1: CGATGAAGTGTACGAGCTGAT, shTIE1-2: GCCACGACCATGACGGCGAAT) were transfected in A549, H1299, and A549 ING5 overexpression cell lines and H1299 ING5 overexpression cell lines.

These cells were grown in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, NY, USA) supplemented with 10% fetal bovine serum (FBS, HyClone, UT, USA), 10 mg/mL antibiotics (penicillin and streptomycin), and 2 mmol/L L-glutamine at 37 °C under 5% CO₂ and saturated moisture.

2.2 Proteomic quantification of protein phosphorylation

Lung cancer A549 cells with ING5 overexpression and A549 control cells were used to quantify dynamic changes of protein phosphorylation in PTM Biolabs (Hangzhou, China). The experiments included SILAC labeling, protein extraction, HPLC fractionation, affinity enrichment, and LC–MS/MS analysis. Intensive bioinformatic analyses were conducted to annotate the quantifiable targets, including protein annotation, GO and KEGG analyses, functional enrichment, and enrichment-based clustering analysis.

2.3 Western blot

Cells were lysed in lysis buffer containing 150 mmol/L NaCl, 1% NP40, 0.5% deoxycholic acid, 0.1% SDS, 50 mmol/L Tris (pH = 8.0), and 1:25 protease inhibitor cocktail for total protein (Thermo Scientific, MA, USA). Protein concentration was determined by using the Bradford protein assay system (Bio-Rad, Hercules, CA). Equal amounts of protein (30 μg of protein each lane) were separated by SDS-PAGE and transferred to nitrocellulose membranes (Hybond-C, Amersham, UK). Immunoblots were blocked with 5% skim milk (Yili Industrial Group Company, Inner Mongolia, China) in PBS/Tween-20 (0.05%, v/v) for 1 h at room temperature. The membrane was incubated with a primary antibody overnight at 4 °C. The primary antibodies used include anti-ING5 and PDK1 antibodies (Proteintech Group Inc., MI, USA); anti-PDHA1, p-PDHA1 (S293), LDHA, HIF-1α, E-cadherin, N-cadherin, TIE1, and GAPDH antibodies (Abcam, Cambridge, UK); anti-p-PDK1 antibody (Y376, Affinity Bioscience, China); and anti-β-actin antibody (Sigma-Aldrich, MO, USA). The p-PDK1 (Y163) antibody was custom-made by Jingjie PTM Biolabs Co., Ltd. (Hangzhou, China). The membrane was incubated with a corresponding secondary antibody conjugated with horseradish peroxidase (Sigma, 1:5000) at room temperature for 1 h. The blots were developed using an enhanced chemiluminescence Western blot detection system (Amersham Bioscience, Buckinghamshire, UK).

Preparation of the custom-made p-PDK1 antibody (Y163): (1) The experimental group (positive polypeptide/negative polypeptide) and control group were set up; three EP tubes were prepared; 2 mL of 5% skimmed milk powder (primary antibody dilution) was added to each tube, and the primary antibody was added to be detected in each tube. (2) Peptides were added to the positive peptide and negative peptide tubes with antibody mass 2.5 times higher than the peptide in accordance with the molar ratio calculation of the excess peptide added. The control group was not added with peptide, only antibody. (3) The tubes were placed on a shaker and incubated for 2 h at room temperature with rotation. (4) The incubated product was added as the primary antibody to the same membrane for routine primary antibody incubation. Follow-up assay was consistent with routine Western blot.

2.4 Immunohistochemical (IHC) staining and tissue microarray (TMA) analysis

The TMAs used in this study were obtained from Outdo Biotech Company (HLugA180Su 07, Shanghai, China). All tissues included in the present study were obtained from 98 patients with lung cancer who are undergoing complete surgical resection of the lung tumor as an initial treatment between July 2004 and June 2009. The specimens were fixed in 10% formalin. Paraffin-embedded tumor specimens and paired adjacent non-tumor specimens were collected for histological examination. All patients were followed up until August 2014. The TMAs contained 98 lung adenocarcinoma tissues and 82 adjacent lung noncancerous tissues. TMA slides were routinely deparaffinized and rehydrated. Antigen retrieval was performed by heating the samples at 95 °C for 30 min in 10 mmol/L sodium citrate (pH = 6.0). Endogenous peroxidase activity was blocked by immersing the samples in 3% hydrogen peroxide for 5 min. Then, the non-specific antibody binding sites were blocked by incubating in 5% bovine serum albumin (BSA) for 30 min. Samples were incubated with the anti-p-PDK1 (Y163) antibody (1:50 dilution, PTM Biolabs, Hangzhou, China) at 4 °C overnight and then with a biotin-labeled secondary antibody followed by streptavidin–peroxidase. The color reaction was developed in 3,3′-diaminobenzidine for 5 min and counterstained with hematoxylin.

2.5 Evaluation of IHC staining

The intensity of p-PDK1-Y163 immunostaining was scaled as no stain (0), weak staining (1+), moderate staining (2+), or strong staining (3+). The percentage of positive cells was scored as follows: 0, 0–5%; 1, 6%–25%; 2, 26%–50%; 3, 51%–75%; 4, 76%–100%. A final score was obtained by multiplying both intensities. Scores of 0 to 6 were considered low p-PDK1 (Y163) expression, whereas scores of 7 to 12 were considered high expression.

2.6 PDH activity assay

Protein extraction and concentration detection of lung cancer cells were described as before. Protein concentration was adjusted to 15 mg/mL with lysis, and the final concentration of lung cancer A549 cells and colon cancer HCT116 cells were adjusted to 700 μg per 200 μL and 300 μg per 200 μL with PBS, respectively. The relevant experimental procedure was performed in accordance with the instructions of the PDH Enzyme Activity Microplate Assay Kit (Sigma-Aldrich, MO, USA). In brief, protein samples were loaded onto 96-well culture plates, including positive control samples and buffer control as a null reference. The plates were incubated for 3 h at room temperature. Then, the wells were rinsed two times with 1× buffer before adding 200 μL of assay solution. The absorbance of each well was measured at 450 nmol/L at 20-s intervals using a kinetic microplate reader (iMarkTM, Bio-Rad, Hercules, CA) for approximately 30 min. The experiments were conducted in triplicate.

2.7 Oxygen consumption assay

Cells were seeded in triplicate in a black, clear-bottom 96-well culture plate at a density of 8 × 10⁴ cells/200 μL per well for A549 cells and 6 × 10⁴ cells/200 μL per well for HCT116 cells. The cells were incubated for 24 h. The experimental procedure was performed in accordance with the instructions of the Oxygen Consumption/Glycolysis Dual Assay Kit (Cayman Chemical Company, MI, USA). In brief, 10 μL of antimycin A stock solution and 10 μL of MitoXpress-Xtra solution were added and then gently overlaid in each well with 100 μL of HS mineral oil. The absorbance was measured kinetically at 37 °C using a variable temperature-controlled microplate reader (Tecan Infinite 200 Pro) for ≥ 120 min at an excitation wavelength of 380 nmol/L and an emission wavelength of 650 nm. The experiments were conducted in triplicate.

2.8 LDH assay

Cell protein was extracted, and the concentration was detected as described previously. The final protein concentration of lung cancer A549 cells and colon cancer HCT116 cells was adjusted to 3 and 0.6 μg/μL with lysis buffer, respectively. The relevant experimental procedure was performed in accordance with the instructions of the LDH Enzyme Activity Microplate Assay Kit (Abcam, Cambridge, UK). In brief, samples were added with 250 μL matrix buffer and 50 μL coenzyme I reaction solution and incubated at 37 °C for 15 min. Then, 250 μL 2,4-dinitrotoluene was added, and the samples were incubated at 37 °C for 15 min. Finally, the samples were added with 2.5 mL of 0.4 mol/L NaOH solution. The absorbance at 450 nm was measured immediately after 3 min of incubation at room temperature using a microplate reader (iMarkTM, Bio-Rad, Hercules, CA). The experiments were conducted in triplicate.

2.9 LD assay

Cells were seeded in triplicate in six-well culture plates at a density of 4 × 10⁵ cells/1 mL per well and incubated in a humidified incubator at 37 °C with 5% CO₂. After incubating the cells for 24 h, cell supernatant was collected as tested samples, and the assay was performed in accordance with the instructions of the lactic acid assay kit (LD, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). In brief, each sample was added with 1 mL of enzyme working solution and 200 μL chromogenic agent before incubation at 37 °C for 10 min. The absorbance was immediately measured after adding a terminal solution at 530 nm using a microplate reader (iMarkTM, Bio-Rad, Hercules, CA). The experiments were conducted in triplicate.

2.10 Proliferation assay

Cells were seeded in triplicate in 96-well culture plates at a density of 2 × 10³ cells/200 µL per well. MTT assay was performed as described previously [12]. Experiments were conducted in triplicate and analyzed by using a paired t-test.

2.11 Colony formation assay

Cells were seeded in triplicate in six-well culture plates at a density of 3 × 10² cells/2 mL per well and incubated in a humidified incubator at 37 °C with 5% CO₂ for 15 days when colonies were visible. Crystal violet staining was performed, and the number of colonies was counted.

2.12 Wound-healing assay

Cells were seeded in six-well plates at a density of 5 × 10⁵ cells/2 mL/well. Once the cells reached about 90% confluence, a wound area was carefully created by scraping the cell monolayer with a sterile 200 μL pipette tip from one end to the other end of the well. The detached cells were removed by washing them with PBS. Cells that migrated to the wounded region were observed by using an Olympus CK-2 inverted microscope and photographed (100× magnification) at 0 h, 8 h, 16 h, 20 h, and 24 h. The experiments were performed in triplicate.

2.13 Transwell migration and invasion assay

For the migration assay, 5 × 10⁴ cells were suspended in a serum-free medium and plated on chambers (Corning Costar, NY, USA) that were not coated with Matrigel. For the invasion assay, the upper chamber was precoated with Matrigel (BD Bioscience, CA, USA) in accordance with the manufacturer’s protocols before 5 × 10⁴ cells in serum-free DMEM were added to the chamber. The assays were performed as described previously.

2.14 Xenograft studies

Male athymic BALB/c nude mice (5–6 weeks old) were bought from the Experimental Animal Center of Fourth Military Medical University. All animal procedures were performed in accordance with the protocols approved by the Animal Care and Use Committee of Fourth Military Medical University (approval no: 20160505). Mice were randomly divided into four groups (n = 6) and then injected subcutaneously with 5 × 10⁶ A549 control cells, ING5-overexpressing cells, ING5-PDK1 WT cells, and ING5-PDK1 Y163F cells. The tumor size was monitored using calipers two times a week. The tumor volume was estimated as (a × b²)/2, where “a” is the large diameter, and “b” is the small diameter of the tumor. The survival time of nude mice was observed and recorded. For the intravenous mouse model, mice were randomly divided into four groups (n = 6) and injected with the abovementioned 5 × 10⁶ lung cancer cells through the tail vein. Mice were sacrificed at day 45 after injection, and the lungs were examined for tumor metastasis. The pathological examination of lung tumor tissues in nude mice was performed.

2.15 Hematoxylin–eosin (HE) staining

HE staining analysis of tissue histological structures was performed using a HE staining kit (Beyotime, C0105S). Tissues were formaldehyde fixed, dehydrated, and embedded in paraffin. The sections were dehydrated with graded ethanol. A light microscope (Olympus) was used to observe the pathological changes.

2.16 High-throughput siRNA screening combined with high-content functional analysis

The human tyrosine kinase siRNA library was obtained from Ribobio (Guangzhou, China), consisting of 91 independent siRNA pools against each gene of tyrosine kinase. ING5-overexpressing A549 cells were seeded in 96-well plates (Corning Costar, NY, USA). siRNA or siRNA-NC were transfected with lipofectamine 3000 (Invitrogen, CA, USA). The upper chamber was precoated with Matrigel (Corning Costar, NY, USA) before 100 μL mix suspension with 2% FBS (Gibco, NY, USA) DMEM was added. The bottom chamber contained 150 μL DMEM with 20% FBS. After incubation for 6 days, the migrated cells were fixed and stained with Calcein AM (2.5 μmol/L, Corning Costar, NY, USA) for 0.5 h. The cells were imaged by using the IN Cell Analyzer 6500HS with a 10× objective for 12 parts per hole. Fluorescence was measured by excitation at 488 nm. Invasion ratio = the number of cells invasion in the siRNA group/number of cells invasion in the siRNA-NC group.

2.17 Analysis of the Gene Expression Profiling Interactive Analysis 2 (GEPIA2), Tumor Immune Estimation Resource (TIMER), and Kaplan–Meier (K–M) plotter databases

The GEPIA2 database is a well-developed web server for analyzing the RNA sequencing expression data based on tumors and normal samples from the Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression databases [27]. TIE1 expression levels of various types of cancer and adjacent normal tissue were obtained by using the GEPIA2 database. Correlation analysis between a pair of genes in a given cancer type was performed using the TIMER [28]. Spearman’s value was used to estimate statistical significance. The correlation between the expression level of TIE1 and ING5 was investigated. The K–M plotter database could assess almost 54 000 survival genes in lung cancer [29]. Samples from patients with lung adenocarcinoma (LUAD) with negative surgical margins were classified into two groups based on the median of TIE1 (Affy ID: 204468) mRNA expression in the K–M plotter database for exhibiting the overall survival (OS).

2.18 In vitro kinase assay

For the TIE1 kinase assay, 1 μg of the purified GST-PDK1 WT (MERRYBIO, China) or GST-PDK1 (Y163F, MERRYBIO, China) recombinant proteins was incubated with 1 μg of recombinant active TIE1 (His-TIE1, Sino Biological Inc., Beijing, China) and 200 μmol/L ATP (Cell Signaling Technology, Inc., MA, USA) in 50 μL of 1× kinase buffer (Cell Signaling Technology, Inc., MA, USA) for 1 h at 30 °C. Reactions were stopped by boiling in the SDS sample buffer, and proteins were resolved by 10% SDS-polyacrylamide gel electrophoresis. After being transferred to a nitrocellulose membrane, the samples were incubated with specific primary antibodies, including anti-phospho-PDK1 (Y163), TIE1, and PDK1.

2.19 Immunofluorescence (IF) assay

Cells were seeded on glass coverslips and then washed in PBS, followed by fixing in 4% formaldehyde for 15 min and then staining with Mito-tracker Red CMXRos (Invitrogen, CA, USA) for mitochondrial imaging Subsequently, cells were incubated with the TIE1 antibody (Abcam, Cambridge, UK) at 4 °C overnight and then with goat anti-rabbit IgG. Nuclei were stained with DAPI. Finally, the cells were imaged on Nikon C2 Plus confocal laser system coupled with a Nikon Eclipse Ti microscope.

2.20 Immunogold transmission electron microscopy (TEM)

A549 cells were obtained by centrifugation at 4 °C using a high-speed refrigerated centrifuge. The precipitate was dehydrated and infiltrated with resin. The resin-containing capsules and samples were moved into a low-temperature UV polymerizer (Electron Microscopy Sciences, PA, USA) and polymerized at −20 °C for more than 48 h. Subsequently, ultrathin sections (70–80 nm) were cut by using an ultra-microtome (Leica, Solms, DE) and mounted on parlodion-coated 150-mesh nickel grids with a formvar film (Zhongjing Keyi, Beijing, China). The section was incubated with 1% BSA TBS solution with TIE1 antibody (1:20) overnight at 4 °C. The nickel grids were incubated in the secondary antibody diluted in a diluent (1:50, Sigma-Aldrich, MO, USA). The nickel grids were observed through TEM (Hitachi, Tokyo, Japan), and images were taken. Furthermore, 10 nm black golden particles indicate positive signals.

2.21 Statistical analysis

Statistical tests were run using Graph Pad Prism 6 (GraphPad Software, CA, USA). Two groups of data were analyzed by using t-test. Three or more samples were compared by one-way ANOVA, followed by Dunnett multiple-comparison test. The survival curves of different groups were analyzed by using the log-rank (Mantel–Cox) test. All statistical tests were two-sided. For all tests, P < 0.05 was regarded as statistically significant.

3 Results

3.1 Quantitative analysis of phospho-proteome by ING5 overexpression in A549 cells

By quantitative phospho-proteomics, 3647 phosphorylation sites were identified from 1436 proteins, and 3561 phosphorylated peptides could be quantified on 1405 proteins. Table S1 shows an excel file comprising information regarding the phosphorylated peptides and proteins.

Quantified phosphorylated proteins were divided into four quantiles in accordance with previously described methods [12]: Q1 (0–15%), Q2 (15%–50%), Q3 (50%–85%), and Q4 (85%–100%, Fig.1). The overall phosphorylation level was upregulated. The phosphorylation of proteins in Q4 (534 phospho-peptides) and the de-phosphorylation of proteins in Q1 (535 phospho-peptides) were promoted by ING5. Although the amounts of phospho-peptides in Q4 and Q1 were almost equal, 411 showed a more than threefold increase in abundance in Q4, whereas only 46 showed a threefold decrease in Q1.

Fig.1 Quantitative analysis of phospho-proteome upon ING5 overexpression. (A) Quantitative ratio (L/H) distribution of phosphorylated protein influenced by ING5 overexpression. The figure shows the protein ratio (log (protein ratio), with base = 2) of each differentially expressed protein. Proteins with upregulated phosphorylation are located right of zero of the x-axis, whereas proteins with downregulated phosphorylation are located left of zero of the x-axis. (B) Enrichment and clustering analysis of biological processes based on Gene Ontology annotation. (C) Enrichment and clustering analysis of molecular function based on Gene Ontology annotation. (D) Enrichment and clustering analysis of cellular components based on Gene Ontology annotation. (E) Enrichment and clustering analysis of KEGG pathways. (F) Enrichment and clustering analysis of protein domain. (G) ING5 participated in the modified protein interaction network of the pathway map05230 central carbon metabolism in cancer. The circle represents the protein; the size of the circle represents the number of interacting proteins, and the different colors represent the Q1–Q4 proteins. (H) Difference ratio information of all phosphorylation sites. The horizontal coordinate is the ratio value rank, and the vertical coordinate is the log2 ratio value. The Q1, Q2, Q3, and Q4 sites are distinguished by color, and the PDK1-Y163 site is separately marked.

Full size|PPT slide

To obtain a comprehensive understanding of the biological relevance of ING5-regulated phosphorylation, GO functional enrichment analysis was performed. Our data showed that ING5-promoted phospho-proteins were significantly enriched during biological processes, including chromatin organization, chromosome organization, chromatin assembly or disassembly, and chromatin modification, whereas ING5-downregulated phospho-proteins were enriched during cytoskeleton organization and carbohydrate metabolic process (Fig.1). Analysis of molecular function showed that ING5-increased phospho-proteins were enriched during DNA binding and protein binding, whereas ING5-decreased phospho-proteins were enriched in isomerase activity (Fig.1). Compartment analysis showed that ING5-increased phospho-proteins were enriched in chromosome, chromatin, nucleus, etc., whereas ING5-decreased phospho-proteins were enriched in cytoskeleton and non-membrane-bounded organelle (Fig.1). Enrichment analysis of KEGG pathways showed that ING5-upregulated phospho-proteins were enriched during lysine degradation (Fig.1). Domain analysis showed that ING5-increased phospho-peptides were enriched in domains of zinc finger (FYVE/PHD-type, PHD-type, and PHD-finger), PDZ, BRCT, etc., whereas ING5-downregulated phospho-peptides were enriched in the SAP domain (Fig.1). By comparing with the STRING (V11.0) protein interaction network database, the protein interaction relationship was extracted in accordance with the confidence score of > 0.4. Network image beautification is achieved using cytoscape3.10.0. ING5, PDK1, and the modified protein interaction network are involved in the pathway map05230 central carbon metabolism in cancer (Fig.1). The PDK1-Y163 site in Q4 was considered to be promoted by ING5 (Fig.1).

3.2 ING5 upregulated the phosphorylation of PDK1 at Y163, which positively correlated with better prognosis in patients with lung cancer

Among ING5-regulated phosphorylation sites, we focused on PDK1 Y163, a new phosphorylation site that is upregulated by 13.847-fold in ING5 overexpression lung cancer A549 cells (Fig.2). Using a phospho-specific antibody against phosphor-PDK1 Y163, we confirmed that PDK1 Y163 phosphorylation was upregulated by ING5 overexpression but downregulated by ING5 knockdown in A549 and H1299 lung cancer cells. No significant change in the expression of PDK1 Y376 was found in the abovementioned cell lines (Fig.2 and 2C).

Fig.2 Y163 is a new phosphorylation site in PDK1 upregulated by ING5 overexpression, which positively correlated with better prognosis in patients with lung cancer. (A) LC–MS/MS spectrum of peptide HNDVIPTMAQGVIEYK of PDK1. (B) Phosphorylation of PDK1 Y163 is upregulated in ING5-overexpressing lung cancer cells by Western blot. Phosphorylation of PDK1 Y376 shows no significant change between the control and high-ING5-expression group. Actin was used as an internal loading control. (C) Phosphorylation of PDK1 Y163 is downregulated in ING5-knockdown lung cancer cells. Phosphorylation of PDK1 Y376 shows no significant change between the control and ING5 knockdown groups. (D, E) Representative images of NSCLC and adjacent noncancerous tissues with p-PDK1 Y163 staining. Scale bar, 200 μmol/L and 50 μmol/L. (F) The expression of p-PDK1 Y163 in NSCLC was lower than that in adjacent noncancerous specific antibody (P < 0.001). (G) Kaplan–Meier overall survival (OS) analysis based on NSCLC patients with high and low p-PDK1 Y163 levels (HR = 0.596, 95% CI = 0.396–0.979, P = 0.041, log-rank test).

Full size|PPT slide

To investigate the clinical significance of PDK1 Y163 phosphorylation, the expression of p-PDK1 Y163 in TMAs was analyzed using IHC staining. The TMAs contained 180 samples (NSCLC: 98, adjacent non-tumor tissue: 82), and positive p-PDK1 Y163 staining was mainly located in the cytoplasm (Fig.2 and 2E). The p-PDK1 Y163 positive rate in NSCLC tissues was 57.1% (56/98) and 93.9% (77/82) in non-tumor tissues. The level of p-PDK1 Y163 in NSCLC was significantly lower than that in non-tumor tissues (7.56 vs. 10.18, P < 0.001, Fig.2). The levels of p-PDK1 Y163 correlated significantly with NSCLC tumor invasion (P = 0.029, Tab.1). Survival analysis was performed for 87 cases of NSCLC. The other 12 NSCLC cases were excluded because of a lack of follow-up information. The results indicated that elevated p-PDK1 Y163 predicted better patient survival (P = 0.041, log-rank test, Fig.2).

Tab.1 Correlation of p-PDK1 Y163 expression with clinicopathological features of NSCLC

Clinicopathological variables	Total (n)	p-PDK1 (Y163) expression (IHC)
		Low	High	χ2	P value
Age
< 60	45	20	25	0.085	0.771
≥ 60	53	22	31
Gender
Male	55	21	34	1.107	0.293
Female	43	21	22
Tumor location
Left lung	43	18	25	0.005	0.943
Right lung	54	23	31
Tumor invasion
T1	20	10	10	9.033	0.029*
T2	51	17	34
T3	22	10	12
T4	5	5	0
Lymph node metastases
N0	44	16	28	1.139	0.286
N1–N3	53	25	28
Metastasis
No	97	55	42	0.750	0.387
Yes	1	1	0
Clinical stage
I–II	52	19	33	2.398	0.122
III–IV	45	24	22
Pathological grade
Well and moderate	84	38	46	1.347	0.246
Poorly and not	14	4	10

Note: *Statistically significant (P < 0.05).

3.3 ING5 overexpression reverses the Warburg effect by upregulating PDK1 Y163 phosphorylation, which negatively regulates PDK1 kinase activity toward PDHA1

PDK1 kinase activity could be activated by phosphorylation at Y136, Y243, and Y244 [26], leading to phosphorylation and inactivation of its target PDHA1 at S293 [23]. However, the mechanism by which Y163 phosphorylation affects PDK1 activity remains unknown. Interestingly, phospho-proteomics showed 3.0-fold downregulation of PDHA1 S293 phosphorylation in ING5-overexpressing A549 cells (Fig.3). Western blot verified that in A549 and H1299 cells, the dephosphorylation of PDHA1 S293 was induced by ING5 overexpression (Fig.3). The decreased phosphorylation of PDHA1 S293 was accompanied with increased enzyme activity of the dehydrogenation of pyruvic acid (Fig.3).

Fig.3 Y163 phosphorylation negatively regulates PDK1 kinase activity, and it is necessary for the ING5-reversed Warburg effect. (A) LC–MS/MS spectrum of peptide YHGHSMSDPGVSTR of PDHA1. Phosphorylation of PDHA1 S293 is downregulated by ING5 overexpression by quantitative phospho-proteomics. (B) ING5 overexpression decreased the phosphorylation of PDHA1 S293 by Western blot. Actin was used as an internal loading control. (C) ING5 overexpression (ING5) increased the enzyme activity of PDH. (D) ING5 overexpression enhanced cancer cell oxygen consumption. (E) ING5 overexpression downregulated HIF-1α and LDHA by Western blot. Actin was used as an internal loading control. (F) ING5 overexpression inhibited LDH activity. (G) ING5 overexpression decreased cancer cell lactic acid (LD) production. (H) Phosphorylation of PDK1 Y163 was decreased by Y163F mutation. ING5 overexpression—decreased PDHA1 S293 phosphorylation was impaired by PDK1 Y163F mutation. Actin was used as an internal loading control. (I) ING5 overexpression—increased PDH enzyme activity was abolished by PDK1 Y163F mutation. (J) ING5 overexpression—enhanced oxygen consumption was decreased by PDK1 Y163F mutation. (K) PDK1 Y163F phospho-dead mutation impaired the downregulating effects of ING5 overexpression on HIF-1α and LDHA levels. (L) ING5 overexpression—downregulated LDH activity was impaired by PDK1 Y163F mutation. (M) ING5 overexpression–decreased lactic acid production was impaired by PDK1 Y163F mutation. Data of (C, D, F, G, I, J, L, and M) are shown as the mean plus standard error of three independent experiments (n = 3 samples). Statistical analysis was performed by one-way ANOVA. *P < 0.05 and **P < 0.01 versus the NC group.

Full size|PPT slide

The key point of the Warburg effect is aerobic glycolysis, which is characterized by decreased oxygen consumption and increased lactate production [13]. Our results showed that ING5 overexpression enhanced the oxygen consumption (Fig.3) of lung cancer cells, indicating increased OXPHOS. Hypoxia-inducible factor-1 alpha (HIF-1α), a transcription factor induced by low-oxygen conditions, plays a critical role in the Warburg effect by upregulating glycolytic enzymes, such as lactate dehydrogenase A (LDHA) [19]. Herein, the results indicated that ING5 overexpression downregulated HIF-1α and LDHA levels (Fig.3). ING5 also inhibited LDH activity (Fig.3), resulting in decreased lactate production (Fig.3), reflecting decreased glycolysis. Thus, the overexpression of ING5 promotes the phosphorylation and inactivation of PDK1 to reverse the Warburg effect in lung cancer cells. Similar results were observed in ING5-overexpressing HCT116 colorectal cancer cells (Fig. S1A– S1G).

3.4 PDK1 Y163 phosphorylation is necessary for the ING5-reversed Warburg effect

The Western blot results indicated that the phosphorylation of PDK1 Y163 in ING5 OE cells and PDK1 WT cells was significantly promoted, with decreased phosphorylation of PDHA1 S293, compared with control cells. However, p-PDK1 (Y163) in PDK1 Y163F cells was decreased, which was accompanied with increased p-PDHA1 (S293, Fig.3). Accordingly, PDH enzyme activity was decreased in PDK1 Y163F mutant cells but not in PDK1 WT cells in the presence of ING5 overexpression (Fig.3). ING5 overexpression enhanced oxygen consumption, which was impaired by PDK1 Y163F mutation (Fig.3). Y163 phospho-dead mutant also abolished the downregulating effect of ING5 overexpression on HIF-1α and LDHA levels (Fig.3), with increased enzyme activity of LDHA (Fig.3) and subsequent high lactate production (Fig.3). These results demonstrated that PDK1 Y163 phosphorylation correlates inversely with its kinase activity toward PDHA1 and plays an important role in the reversal of the Warburg effect induced by ING5 overexpression.

3.5 PDK1 Y163 mutation impairs the ING5—inhibited proliferation and invasion of lung cancer cells

To define the role of PDK1 Y163 phosphorylation in ING5-reversed cancer cell metabolism reprogramming, the following cell lines were used: (1) control A549 cells (control); (2) ING5-overexpressing A549 cells (ING5 OE); (3) ING5-overexpressing A549 cells with PDK1 knockdown (PDK1 KD); (4) ING5 OE-PDK1 KD cells rescued with wildtype PDK1 (PDK1 WT); (5) ING5 OE-PDK1 KD cells rescued with phospho-dead Y163F mutant PDK1 (PDK1 Y163F). The effects of PDK1 Y163 mutation on the anti-proliferative and anti-invasive functions of ING5 overexpression were assessed. The results showed that PDK1 Y163 mutation resulted in a significantly increased proliferation (Fig.4) and colony formation of ING5-overexpressing A549 cells (Fig.4). Wound healing and Transwell migration assays showed that PDK1 Y163 mutation promoted the migration of ING5-overexpressing A549 cells (Fig.4 and 4D). PDK1 Y163 mutation also significantly enhanced the invasion of ING5-overexpressing A549, as assessed using matrigel-coated Transwell chambers (Fig.4). In addition, PDK1 Y163F impaired the inhibitory effects of ING5 on EMT by downregulating E-cadherin while upregulating N-cadherin (Fig.4). Collectively, the phosphorylation of PDK1 at Y163 is an important step in the inhibition of EMT and cancer invasion induced by ING5 overexpression.

Fig.4 PDK1 Y163 phosphorylation is required for ING5-inhibited invasiveness of lung cancer cells in vitro and in vivo. Cells were divided into the following groups: (1) control A549 cells (control); (2) ING5-overexpressing cells (ING5); (3) ING5-overexpressing A549 cells with PDK1 knockdown (ING5-PDK1 KD); (4) ING5-PDK1 KD cells rescued with wild-type PDK1 (ING5-PDK1 WT); (5) ING5-PDK1 KD cells rescued with phospho-dead Y163F mutant PDK1 (ING5-PDK1 Y163F). (A) PDK1 Y163F mutation significantly increased the proliferation of lung cancer A549 cells with ING5 overexpression. (B) PDK1 Y163F mutation significantly increased colony formation of lung cancer A549 cells with ING5 overexpression. Representative pictures are shown. Colony numbers were quantified. (C) Wound-healing assay was performed to show the promoting effects of PDK1 Y163 mutation on the migration of A549 ING5-overexpressing cells. A scratch wound was made on the cell surface, and cells were photographed at 0 h, 8 h, 16 h, 20 h, and 24 h. Representative pictures are shown. (D) PDK1 Y163 mutation accelerated the Transwell migration of A549 ING5-overexpressing cells. The migrated cells were photographed (100× magnification). Representative pictures are shown. (E) PDK1 Y163 mutation promoted the Transwell invasive abilities of A549 ING5-overexpressing cells. The invaded cells were photographed (100× magnification). Representative pictures are shown. (F) Effects of PDK1 Y163 mutation on the expression of EMT markers in A549 ING5-overexpressing cells. Actin was used as an internal loading control. (G) Mice were injected subcutaneously with 5 × 106 lung cancer A549 cells. At the end of the experiment, mice were sacrificed and photographed. (H) Tumor growth curves revealed that the tumor volumes were significantly higher in mice with ING5-PDK1 Y163F cells. (I) Survival analysis showed that mice with ING5 OE cells had better overall survival than those with control cells (P = 0.0007), whereas mice with ING5-PDK1 Y163F cells had worse overall survival than those with ING5-PDK1 WT cells (P = 0.0023). (J) Mice were injected through the tail vein with 5 × 106 lung cancer A549 cells. On day 45 after tumor cell injection, mice were sacrificed, and the lungs were photographed. Gross images of the lungs showed lung-metastasized tumors in different groups. (K) Representative images of HE staining for lung tissues were provided (scale bar, 200 μm). (L) Tumor index of mice from different groups. (M) Lung weight of mice from different groups. Statistical analysis was performed using two-way ANOVA, followed by post-hoc tests. *P < 0.05 and **P < 0.01 versus the NC group.

Full size|PPT slide

3.6 PDK1 Y163 phosphorylation contributes to ING5 overexpression—inhibited cancer growth and metastasis in mouse xenograft models

A subcutaneous mouse xenograft model was constructed by inoculating nude mice with control, ING5 OE, PDK1 WT, and PDK1 Y163F A549 cells to assess whether ING5 overexpression could inhibit the growth of lung cancer in vivo by promoting PDK1 Y163 phosphorylation. The tumor volume was measured, and the survival time of nude mice was observed. ING5 overexpression significantly inhibited tumor growth, with decreased tumor volumes compared with those in the control group. However, the inhibitory effects of ING5 were almost completely abolished by PDK1 Y163F mutation (Fig.4 and 4H). Furthermore, the OS of mice injected with ING5 OE cells was longer than that of the control mice, whereas mice injected with PDK1 Y163F A549 cells had significantly shorter lifetimes compared with those in the PDK1 WT group (Fig.4). These results demonstrated that PDK1 Y163 phosphorylation contributes to ING5-induced growth inhibition and longer survival.

Next, an intravenous mouse xenograft model was made by tail vein injection of lung cancer cells in nude mice to validate the role of PDK1 Y163 phosphorylation in the anti-metastatic effects of ING5 in vivo. On day 45 post-injection, all five mice injected with control A549 cells had developed multiple tumors in their bilateral lungs. By contrast, the mice in the ING5 OE group or PDK1 WT group developed fewer lung tumors than those in the control group (Fig.4 and 4K). However, the mice in the PDK1 Y163F group developed more lung tumors than those in the PDK1 WT group (Fig.4 and 4K). A previously described grading system was used to calculate the tumor index [30]. Compared with the mice in the PDK1 WT group, those in the PDK1 Y163F group had significantly higher tumor index and lung weights (Fig.4 and 4M). Collectively, these results indicated that PDK1 Y163 phosphorylation is important for ING5-inhibited lung cancer metastasis in vivo.

3.7 Screening with a TPK siRNA library combined with high-content functional analysis identified TIE1 as an anti-tumor TPK upregulated by ING5

To investigate the mechanisms mediating the upregulation of PDK1 Y163 phosphorylation by ING5, we performed a TPK siRNA library (n = 91)-based screening invasion assay with ING5-overexpressing A549 cells. The results for the invasion ratio are shown in Fig. S2 and Fig. S3. Representative images of the invasion results for the top 10 targeted genes are shown in Fig.5, whose knockdown caused the increased invasion of ING5-overexpressing A549 cells. We selected the top three kinase spleen-associated tyrosine kinase (SYK), TIE1, and TEK (TIE2) for further screening.

Fig.5 TIE1 was identified as an anti-tumor TPK upregulated by ING5. (A) Results of the top 10 targeted genes were shown from a TPK siRNA library-based screening of invasion assay. (B) TIE1 mRNA expression in diverse tumor and normal tissues using the GEPIA2 database. Red, tumor samples; green, normal control sample. TIE1 expression was decreased in 11 tumors: LUAD (lung adenocarcinoma), BLCA (bladder urothelial carcinoma), BRCA (breast invasive carcinoma), CESC (cervical squamous cell carcinoma and endocervical adenocarcinoma), COAD (colon adenocarcinoma), KICH (kidney chromophobe), KIRP (kidney renal papillary cell carcinoma), LUSC (lung squamous cell carcinoma), THCA (thyroid carcinoma), UCEC (uterine corpus endometrial carcinoma), and UCS (uterine carcinosarcoma). (C) The mRNA expression level of TIE1 was analyzed in LUAD and normal tissues using TCGA samples. *P < 0.05, compared with normal tissues. (D) TIE1 protein expression levels in LUAD and normal tissue were analyzed using CPTAC samples. (E) Kaplan–Meier overall survival curves showed that the high expression level of TIE1 predicts longer survival. (F) Spearman correlation analysis showed that the expression of TIE1 was positively correlated with ING5 expression. (G, H) Representative images and results of Transwell migration assays of the TIE1-knockdown (sh-TIE1-1 and sh-TIE1-2) A549 control group and A549-ING5-OE control group. (I, J) Representative images and results of Transwell invasion assays of the TIE1-knockdown (sh-TIE1-1 and sh-TIE1-2) A549 control group and A549-ING5-OE control group. Data of (G−J) are shown as the mean plus standard error of three independent experiments (n = 3 samples).

Full size|PPT slide

First, SYK was excluded because clinical data showed no significant difference in its expression between normal and lung cancer samples (Fig. S4). The GEPIA2 database was used to analyze the differential expression of TIE1 in diverse tumor and normal tissues [27]. Among 33 kinds of malignant tumors analyzed, 11 tumors had significantly downregulated TIE1 levels in tumor tissues (Fig.5). In LUAD, the TIE1 expression was significantly downregulated compared with that in normal control tissues (8.03 in normal tissue vs. 45.7 in lung adenocarcinoma, P < 0.001). In the TCGA database, TIE1 mRNA levels in normal control samples were remarkably higher than those in tumor tissues (Fig.5, P < 0.05). Then, the National Cancer Institute’s Clinical Proteomic Tumor Analysis Consortium database was used to assess the TIE1 protein level in patients with LUAD [31], which showed that the protein level of TIE1 in normal tissue was significantly higher than that in LUAD (Fig.5). The data for TIE1 mRNA expression in patients with LUAD showed that the high expression level of TIE1 (Affymetrix ID: 204468_s_at) predict longer survival compared with low expression in the K–M plotter database (Fig.5; hazard ratio (HR) = 0.44, 95% confidence interval (CI) = 0.20–0.95, P = 0.032) [29]. By contrast, the high expression level of TIE2 (TEK, Affymetrix ID: 217711_at) correlated with poor patient survival (Fig. S5A). Spearman correlation analysis demonstrated a positive correlation between TIE1 expression and ING5 expression (r = 0.133, P = 0.00246) by Spearman’s rho value based on the TIMER database (Fig.5) [28].

To investigate the function of TIE1 in lung cancer, we observed the effects of TIE1 knockdown (sh-TIE1-1 and sh-TIE1-2) on cancer cell invasiveness. The results showed that TIE1 knockdown significantly increased the migratory and invasive ability of control and ING5-overexpressing A549 cells (Fig.5–5J. P < 0.01). By contrast, TIE2 knockdown decreased the migration and invasion of A549 cells (Fig. S5B–S5E). Collectively, TIE1 expression correlated positively with ING5 expression and functioned as a tumor-suppressive TPK in lung cancer.

3.8 TIE1 phosphorylates PDK1 at Y163, which is promoted by ING5

To define whether ING5—promoted PDK1 Y163 phosphorylation is mediated by TIE1, we conducted an in vitro kinase assay. The results showed that recombinant active TIE1 could directly phosphorylate PDK1 Y163, but it failed to phosphorylate PDK1 Y163F (Fig.6). IF assays with MitoTracker demonstrated that TIE1 was localized to the mitochondria, whereas TIE1 did not colocalize with the nuclear marker DAPI in A549 and H1299 cells (Fig.6). In addition, TEM showed that TIE1 localized mainly in the cytoplasm and mitochondria of A549 cells (Fig.6). The knockdown of TIE1 (sh-TIE1-1 and sh-TIE1-2) led to the decreased phosphorylation of PDK1Y163 in lung cancer cells (Fig.6 and 6E). ING5 overexpression upregulated TIE1 levels in lung cancer cells and promoted the phosphorylation of PDK1Y163, whereas ING5 knockdown decreased TIE1 and PDK1Y163 phosphorylation levels (Fig.6 and 6G). A graphic illustration of the proposed mechanism in this study (Fig.6). In brief, ING5 overexpression upregulates TIE1, which phosphorylates PDK1 at Y163, leading to the dephosphorylation-dependent activation of PDHA1 and subsequent increased OXPHOS. ING5 overexpression downregulates HIF-1α/LDHA1, causing decreased glycolysis. By contrast, the TIE2 protein level was upregulated in ING5 knockdown lung cancer cells but downregulated in ING5-overexpressing lung cancer cells (Fig. S6A). TIE2 knockdown led to the increased phosphorylation of PDK1Y163 in lung cancer cells (Fig. S6B). Collectively, ING5 promoted the TIE1-mediated phosphorylation of PDK1 at Y163.

Fig.6 TIE1 directly phosphorylates PDK1 at Y163, which is promoted by ING5. (A) PDK1 Y163 is specifically phosphorylated by rTIE1 by in vitro kinase assay. An antibody that specifically recognizes p-PDK1 Y163 was used. (B) Immunofluorescence assay showed the mitochondrial location of TIE1 in lung cancer cells. Representative images are presented (600×). (C) Mitochondrial localization of TIE1 in A549 cells was confirmed by transmission electron microscopy (TEM). Scale bars: 5.0 μm (left), 1.0 μm (middle), and 500 nm (right). (D, E) ING5 overexpression-promoted PDK1Y163 phosphorylation was impaired by TIE1 knockdown by Western blot. (F, G) Expression of TIE1 and p-PDK1Y163 in ING5-overexpressing or knockdown lung cancer cells by Western blot. (H) A graphic illustration of the proposed mechanism in this study. In brief, ING5 overexpression upregulates TIE1, which phosphorylates PDK1 at Y163, leading to the dephosphorylation-dependent activation of PDHA1 and subsequent increased oxidative phosphorylation. In addition, ING5 overexpression downregulates HIF-1α/LDHA1, thereby decreasing glycolysis.

Full size|PPT slide

4 Discussion

ING5 functions as a tumor suppressor by an as-yet-unknown mechanism. The present study, using quantitative phospho-proteomics, is the first systematic analysis of ING5 overexpression—induced protein phosphorylation. We examined the dynamics of phosphorylation/dephosphorylation of more than 1000 protein sites, which suggested that ING5 might participate in a wide variety of cellular functions through diverse mechanisms. Among the phosphorylated proteins upregulated by ING5, we focused on a new site, Y163, in the metabolic regulator PDK1, which was upregulated by 13.847-fold in ING5-overexpressing A549 cells, ranking the 10th among all the upregulated phospho-sites. These results, together with those of our previous report, indicated that ING5 might participate in the regulation of lung cancer metabolic reprogramming.

Metabolic reprogramming from OXPHOS to glycolysis is a hallmark of cancer. The ratio of glycolysis to OXPHOS in cancer cells determines the degree of cancer aggressiveness [32,33]. Reversing the altered tumor metabolism has shown promise in cancer therapy [34,35]. As a glycolytic enzyme, PDK1 is involved in cancer growth and metastasis [36]. PDK1 expression is elevated and is necessary for efficient liver metastasis in patients with breast cancer [37]. PDK1 also mediates EGFR activation—promoted head and neck squamous cell carcinoma metastasis [38]. PDK1 kinase activity is tightly regulated by tyrosine phosphorylation. Diverse oncogenic tyrosine kinases can phosphorylate and activate PDK1 at Y136, Y243, and Y244, leading to the phosphorylation and inactivation of PDHA1, thereby inducing the Warburg effect and tumor growth [26]. Dichloroacetate, a metabolic inhibitor of PDK1, switches glucose metabolism to OXPHOS from glycolysis [39]. These findings indicate that targeting the PDK1/PDHA1 axis might reverse cancer cell metabolic reprogramming and cancer malignancy. Herein, ING5 overexpression—promoted phosphorylation at Y163 of PDK1 was demonstrated to negatively regulate PDK1 kinase activity toward PDHA1 S293, leading to increased PDHA1 activity and a shift in metabolism from glycolysis to OXPHOS. A phospho-dead mutation at Y163 impaired the inhibitory effects of ING5 on lung cancer cell invasiveness and EMT in vitro as well as lung cancer growth and metastasis in vivo. Furthermore, high levels of PDK1 Y163 phosphorylation in patients with lung cancer correlated with better survival.

Protein tyrosine kinases (PTKs) are involved in various cellular functions, including proliferation, migration, survival, metabolism, and cell cycle control. Genetic changes or dysregulation of PTKs are implicated in a wide range of diseases, including cancer. Oncogenic and anti-cancer tyrosine kinases are found. In the current study, TIE1 was identified as the direct upstream kinase of PDK1 Y163, which could be upregulated by ING5. PTKs can be subdivided into receptor tyrosine kinases (RTKs) and nonreceptor tyrosine kinases. TIE1 is an RTK that is essential for vascular development [40]. Compared with other cell-surface RTKs with an intrinsic, ligand-controlled tyrosine-kinase activity, TIE1 has not been demonstrated to bind directly to any known ligands for activation, and it is still generally considered as an orphan receptor [41]. In endothelial cells, TIE1 physically associates with its structurally related receptor TIE2 (TEK) and downregulates TIE-2-driven signals, including AKT and MAPK phosphorylation, inhibition of apoptosis, and endothelial survival, which are induced by the TIE2 ligand, angiopoietin (Ang) [42,43]. Under hypoxic conditions, TIE1 is critical to reduce Ang1-induced TIE2 activity and angiogenesis [44]. However, the precise function of TIE1 and its mode of action, especially in cancer, remain unclear and controversial. In the current study, we found that TIE1 was also located in the mitochondria, and we demonstrated that ING5-upregulated TIE1 phosphorylates PDK1 at Y163, which represents a negative regulation mechanism of PDK1 kinase activity toward PDHA1 S293. ING5 has been known to form different HAT complexes by associating with HBO1 and MOZ/MORF; thus, ING5 has been defined as a transcription coactivator [2]. However, the precise mechanisms by which ING5 upregulates TIE1 level need to be further investigated.

HIF-1α, which is upregulated in malignant solid tumors, plays a central role in reprogramming cancer metabolism toward aerobic glycolysis by inducing the overexpression and increased activity of several glycolytic enzymes, including PDK1 and LDHA [45–47]. LDHA, which converts pyruvate to lactate, is a key enzyme in aerobic glycolysis, which is overexpressed in many human cancers and correlates with metastasis, tumor recurrence, and poor clinical outcomes [48,49]. Our data showed a significant decrease in the protein levels of HIF-1α and LDHA by ING5 overexpression, although the PDK1 protein level was not particularly affected. These results indicated that ING5 might regulate cancer metabolism reprogramming through multiple mechanisms.

In summary, we identified a novel signaling pathway, ING5/TIE1/p-PDK1 Y163, which could reverse the Warburg effect and inhibit lung cancer invasion and metastasis. Molecular targeting of PDK1 kinase activity by upregulating Y163 phosphorylation shows great application potential as a broadly effective therapeutic strategy.

References

[1] Yuan M, Zhao Y, Arkenau HT, Lao T, Chu L, Xu Q. Signal pathways and precision therapy of small-cell lung cancer. Signal Transduct Target Ther 2022; 7(1): 187–204

[2] Doyon Y, Cayrou C, Ullah M, Landry AJ, Côté V, Selleck W, Lane WS, Tan S, Yang XJ, Côté J. ING tumor suppressor proteins are critical regulators of chromatin acetylation required for genome expression and perpetuation. Mol Cell 2006; 21(1): 51–64

[3] Anwar S, Shamsi A, Mohammad T, Islam A, Hassan MI. Targeting pyruvate dehydrogenase kinase signaling in the development of effective cancer therapy. Biochim Biophys Acta Rev Cancer 2021; 1876(1): 188568–188592

[4] Cui S, Liao X, Ye C, Yin X, Liu M, Hong Y, Yu M, Liu Y, Liang H, Zhang CY, Chen X. ING5 suppresses breast cancer progression and is regulated by miR-24. Mol Cancer 2017; 16(1): 89–100

[5] Ghafouri-Fard S, Taheri M, Baniahmad A. Inhibitor of growth factors regulate cellular senescence. Cancers (Basel) 2022; 14(13): 3107–3118

[6] Archambeau J, Blondel A, Pedeux R. Focus-ING on DNA integrity: implication of ING proteins in cell cycle regulation and DNA repair modulation. Cancers (Basel) 2019; 12(1): 58–76

[7] Wang F, Wang AY, Chesnelong C, Yang Y, Nabbi A, Thalappilly S, Alekseev V, Riabowol K. ING5 activity in self-renewal of glioblastoma stem cells via calcium and follicle stimulating hormone pathways. Oncogene 2018; 37(3): 286–301

[8] Ormaza G, Rodríguez JA, Ibáñez de Opakua A, Merino N, Villate M, Gorroño I, Rábano M, Palmero I, Vilaseca M, Kypta R, Vivanco MDM, Rojas AL, Blanco FJ. The tumor suppressor ING5 is a dimeric, bivalent recognition molecule of the histone H3K4me3 mark. J Mol Biol 2019; 431(12): 2298–2319

[9] Zhang F, Zhang X, Meng J, Zhao Y, Liu X, Liu Y, Wang Y, Li Y, Sun Y, Wang Z, Mei Q, Zhang T. ING5 inhibits cancer aggressiveness via preventing EMT and is a potential prognostic biomarker for lung cancer. Oncotarget 2015; 6(18): 16239–16252

[10] Liu XL, Zhang XT, Meng J, Zhang HF, Zhao Y, Li C, Sun Y, Mei QB, Zhang F, Zhang T. ING5 knockdown enhances migration and invasion of lung cancer cells by inducing EMT via EGFR/PI3K/Akt and IL-6/STAT3 signaling pathways. Oncotarget 2017; 8(33): 54265–54276

[11] Liu XL, Meng J, Zhang XT, Liang XH, Zhang F, Zhao GR, Zhang T. ING5 inhibits lung cancer invasion and epithelial-mesenchymal transition by inhibiting the WNT/β-catenin pathway. Thorac Cancer 2019; 10(4): 848–855

[12] Zhang T, Meng J, Liu X, Zhang X, Peng X, Cheng Z, Zhang F. ING5 differentially regulates protein lysine acetylation and promotes p300 autoacetylation. Oncotarget 2018; 9(2): 1617–1629

[13] Warburg O. On the origin of cancer cells. Science 1956; 123(3191): 309–314

[14] Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011; 144(5): 646–674

[15] Pusapati RV, Daemen A, Wilson C, Sandoval W, Gao M, Haley B, Baudy AR, Hatzivassiliou G, Evangelista M, Settleman J. mTORC1-dependent metabolic reprogramming underlies escape from glycolysis addiction in cancer cells. Cancer Cell 2016; 29(4): 548–562

[16] Erdem A, Marin S, Pereira-Martins DA, Cortés R, Cunningham A, Pruis MG, de Boer B, van den Heuvel FAJ, Geugien M, Wierenga ATJ, Brouwers-Vos AZ, Rego EM, Huls G, Cascante M, Schuringa JJ. The glycolytic gatekeeper PDK1 defines different metabolic states between genetically distinct subtypes of human acute myeloid leukemia. Nat Commun 2022; 13(1): 1105–1120

[17] Lv L, Lei Q. Proteins moonlighting in tumor metabolism and epigenetics. Front Med 2021; 15(3): 383–403

[18] Cenigaonandia-Campillo A, Serna-Blasco R, Gómez-Ocabo L, Solanes-Casado S, Baños-Herraiz N, Puerto-Nevado LD, Cañas JA, Aceñero MJ, García-Foncillas J, AguileraÓ. Vitamin C activates pyruvate dehydrogenase (PDH) targeting the mitochondrial tricarboxylic acid (TCA) cycle in hypoxic KRAS mutant colon cancer. Theranostics 2021; 11(8): 3595–3606

[19] Hong SM, Lee YK, Park I, Kwon SM, Min S, Yoon G. Lactic acidosis caused by repressed lactate dehydrogenase subunit B expression down-regulates mitochondrial oxidative phosphorylation via the pyruvate dehydrogenase (PDH)-PDH kinase axis. J Biol Chem 2019; 294(19): 7810–7820

[20] Adant I, Bird M, Decru B, Windmolders P, Wallays M, de Witte P, Rymen D, Witters P, Vermeersch P, Cassiman D, Ghesquière B. Pyruvate and uridine rescue the metabolic profile of OXPHOS dysfunction. Mol Metab 2022; 63: 101537–101550

[21] Imanaka S, Shigetomi H, Kobayashi H. Reprogramming of glucose metabolism of cumulus cells and oocytes and its therapeutic significance. Reprod Sci 2022; 29(3): 653–667

[22] Seyfried TN, Arismendi-Morillo G, Mukherjee P, Chinopoulos C. On the origin of ATP synthesis in cancer. iScience 2020; 23(11): 101761–101781

[23] Fan J, Shan C, Kang HB, Elf S, Xie J, Tucker M, Gu TL, Aguiar M, Lonning S, Chen H, Mohammadi M, Britton LM, Garcia BA, Alečković M, Kang Y, Kaluz S, Devi N, Van Meir EG, Hitosugi T, Seo JH, Lonial S, Gaddh M, Arellano M, Khoury HJ, Khuri FR, Boggon TJ, Kang S, Chen J. Tyr phosphorylation of PDP1 toggles recruitment between ACAT1 and SIRT3 to regulate the pyruvate dehydrogenase complex. Mol Cell 2014; 53(4): 534–548

[24] De Rosa V, Iommelli F, Terlizzi C, Leggiero E, Camerlingo R, Altobelli GG, Fonti R, Pastore L, Del Vecchio S. Non-canonical role of PDK1 as a negative regulator of apoptosis through macromolecular complexes assembly at the ER-mitochondria interface in oncogene-driven NSCLC. Cancers (Basel) 2021; 13(16): 4133–4146

[25] Calleja V, Laguerre M, de Las Heras-Martinez G, Parker PJ, Requejo-Isidro J, Larijani B. Acute regulation of PDK1 by a complex interplay of molecular switches. Biochem Soc Trans 2014; 42(5): 1435–1440

[26] Hitosugi T, Fan J, Chung TW, Lythgoe K, Wang X, Xie J, Ge Q, Gu TL, Polakiewicz RD, Roesel JL, Chen GZ, Boggon TJ, Lonial S, Fu H, Khuri FR, Kang S, Chen J. Tyrosine phosphorylation of mitochondrial pyruvate dehydrogenase kinase 1 is important for cancer metabolism. Mol Cell 2011; 44(6): 864–877

[27] Tang Z, Li C, Kang B, Gao G, Li C, Zhang Z. GEPIA: a web server for cancer and normal gene expression profiling and interactive analyses. Nucleic Acids Res 2017; 45(W1): W98–W102

[28] Li T, Fan J, Wang B, Traugh N, Chen Q, Liu JS, Li B, Liu XS. TIMER: a web server for comprehensive analysis of tumor-infiltrating immune cells. Cancer Res 2017; 77(21): e108–e110

[29] Győrffy B. Discovery and ranking of the most robust prognostic biomarkers in serous ovarian cancer. Geroscience 2023; 45(3): 1889–1898

[30] Gautam A, Waldrep JC, Densmore CL, Koshkina N, Melton S, Roberts L, Gilbert B, Knight V. Growth inhibition of established B16-F10 lung metastases by sequential aerosol delivery of p53 gene and 9-nitrocamptothecin. Gene Ther 2002; 9(5): 353–357

[31] Chen F, Zhang Y, Chandrashekar DS, Varambally S, Creighton CJ. Global impact of somatic structural variation on the cancer proteome. Nat Commun 2023; 14(1): 5637

[32] Atas E, Oberhuber M, Kenner L. The implications of PDK1-4 on tumor energy metabolism, aggressiveness and therapy resistance. Front Oncol 2020; 10: 583217–583225

[33] Wang T, Chuffart F, Bourova-Flin E, Wang J, Mi J, Rousseaux S, Khochbin S. Histone variants: critical determinants in tumour heterogeneity. Front Med 2019; 13(3): 289–297

[34] Szlosarek PW, Lee S, Pollard PJ. Rewiring mitochondrial pyruvate metabolism: switching off the light in cancer cells. Mol Cell 2014; 56(3): 343–344

[35] Burns JE, Hurst CD, Knowles MA, Phillips RM, Allison SJ. The Warburg effect as a therapeutic target for bladder cancers and intratumoral heterogeneity in associated molecular targets. Cancer Sci 2021; 112(9): 3822–3834

[36] Peng F, Wang JH, Fan WJ, Meng YT, Li MM, Li TT, Cui B, Wang HF, Zhao Y, An F, Guo T, Liu XF, Zhang L, Lv L, Lv DK, Xu LZ, Xie JJ, Lin WX, Lam EW, Xu J, Liu Q. Glycolysis gatekeeper PDK1 reprograms breast cancer stem cells under hypoxia. Oncogene 2018; 37(8): 1062–1074

[37] Dupuy F, Tabariès S, Andrzejewski S, Dong Z, Blagih J, Annis MG, Omeroglu A, Gao D, Leung S, Amir E, Clemons M, Aguilar-Mahecha A, Basik M, Vincent EE, St-Pierre J, Jones RG, Siegel PM. PDK1-dependent metabolic reprogramming dictates metastatic potential in breast cancer. Cell Metab 2015; 22(4): 577–589

[38] Hsu JY, Chang JY, Chang KY, Chang WC, Chen BK. Epidermal growth factor-induced pyruvate dehydrogenase kinase 1 expression enhances head and neck squamous cell carcinoma metastasis via up-regulation of fibronectin. FASEB J 2017; 31(10): 4265–4276

[39] Škorja Milić N, Dolinar K, Miš K, Matkovič U, Bizjak M, Pavlin M, Podbregar M, Pirkmajer S. Suppression of pyruvate dehydrogenase kinase by dichloroacetate in cancer and skeletal muscle cells is isoform specific and partially independent of HIF-1α. Int J Mol Sci 2021; 22(16): 8610–8635

[40] Savant S, La Porta S, Budnik A, Busch K, Hu J, Tisch N, Korn C, Valls AF, Benest AV, Terhardt D, Qu X, Adams RH, Baldwin HS, Ruiz de Almodóvar C, Rodewald HR, Augustin HG. The orphan receptor Tie1 controls angiogenesis and vascular remodeling by differentially regulating Tie2 in tip and stalk cells. Cell Rep 2015; 12(11): 1761–1773

[41] Singhal M, Gengenbacher N, La Porta S, Gehrs S, Shi J, Kamiyama M, Bodenmiller DM, Fischl A, Schieb B, Besemfelder E, Chintharlapalli S, Augustin HG. Preclinical validation of a novel metastasis-inhibiting Tie1 function-blocking antibody. EMBO Mol Med 2020; 12(6): e11164

[42] Zwiers PJ, Jongman RM, Kuiper T, Moser J, Stan RV, Göthert JR, van Meurs M, Popa ER, Molema G. Pattern of tamoxifen-induced Tie2 deletion in endothelial cells in mature blood vessels using endo SCL-Cre-ERT transgenic mice. PLoS One 2022; 17(6): e0268986

[43] Korhonen EA, Murtomäki A, Jha SK, Anisimov A, Pink A, Zhang Y, Stritt S, Liaqat I, Stanczuk L, Alderfer L, Sun Z, Kapiainen E, Singh A, Sultan I, Lantta A, Leppänen VM, Eklund L, He Y, Augustin HG, Vaahtomeri K, Saharinen P, Mäkinen T, Alitalo K. Lymphangiogenesis requires Ang2/Tie/PI3K signaling for VEGFR3 cell-surface expression. J Clin Invest 2022; 132(15): e155478

[44] Yun JH, Lee HM, Lee EH, Park JW, Cho CH. Hypoxia reduces endothelial Ang1-induced Tie2 activity in a Tie1-dependent manner. Biochem Biophys Res Commun 2013; 436(4): 691–697

[45] Chen F, Chen J, Yang L, Liu J, Zhang X, Zhang Y, Tu Q, Yin D, Lin D, Wong PP, Huang D, Xing Y, Zhao J, Li M, Liu Q, Su F, Su S, Song E. Extracellular vesicle-packaged HIF-1α-stabilizing lncRNA from tumour-associated macrophages regulates aerobic glycolysis of breast cancer cells. Nat Cell Biol 2019; 21(4): 498–510

[46] Semenza GL. HIF-1 mediates metabolic responses to intratumoral hypoxia and oncogenic mutations. J Clin Invest 2013; 123(9): 3664–3671

[47] Hulea L, Gravel SP, Morita M, Cargnello M, Uchenunu O, Im YK, Lehuédé C, Ma EH, Leibovitch M, McLaughlan S, Blouin MJ, Parisotto M, Papavasiliou V, Lavoie C, Larsson O, Ohh M, Ferreira T, Greenwood C, Bridon G, Avizonis D, Ferbeyre G, Siegel P, Jones RG, Muller W, Ursini-Siegel J, St-Pierre J, Pollak M, Topisirovic I. Translational and HIF-1α-dependent metabolic reprogramming underpin metabolic plasticity and responses to kinase inhibitors and biguanides. Cell Metab 2018; 28(6): 817–832.e8

[48] Brand A, Singer K, Koehl GE, Kolitzus M, Schoenhammer G, Thiel A, Matos C, Bruss C, Klobuch S, Peter K, Kastenberger M, Bogdan C, Schleicher U, Mackensen A, Ullrich E, Fichtner-Feigl S, Kesselring R, Mack M, Ritter U, Schmid M, Blank C, Dettmer K, Oefner PJ, Hoffmann P, Walenta S, Geissler EK, Pouyssegur J, Villunger A, Steven A, Seliger B, Schreml S, Haferkamp S, Kohl E, Karrer S, Berneburg M, Herr W, Mueller-Klieser W, Renner K, Kreutz M. LDHA-associated lactic acid production blunts tumor immunosurveillance by T and NK cells. Cell Metab 2016; 24(5): 657–671

[49] Mishra D, Banerjee D. Lactate dehydrogenases as metabolic links between tumor and stroma in the tumor microenvironment. Cancers (Basel) 2019; 11(6): 750–770

Acknowledgements

This study was supported by the National Natural Science Foundation of China (Nos. 81672269, 81272587, and 81172223).

Electronic Supplementary Material

Supplementary material is available in the online version of this article at https://doi.org/10.1007/s11684-024-1057-7 and is accessible for authorized users.

Compliance with ethics guidelines

Conflicts of interest Haihua Zhang, Xinli Liu, Junqiang Li, Jin Meng, Wan Huang, Xuan Su, Xutao Zhang, Guizhou Gao, Xiaodong Wang, Haichuan Su, Feng Zhang, and Tao Zhang declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The study was approved by the the Animal Care and Use Committee of Fourth Military Medical University (approval no: 20160505) and the study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.. Informed consent was obtained from all patients for being included in the study.

RIGHTS & PERMISSIONS

2024 Higher Education Press