EGFR-PKM2 signaling promotes the metastatic potential of nasopharyngeal carcinoma through induction of FOSL1 and ANTXR2

Abstract

Nasopharyngeal carcinoma (NPC) is notorious for its aggressiveness and high metastatic potential. NPC patients with distant metastasis have a particularly poor prognosis; however, evaluating metastatic potential by expression profiles of primary tumors is challenging. This study aimed to investigate the association between activation of epidermal growth factor receptor (EGFR) signaling and NPC metastasis and the underlying mechanisms. We found an association between EGFR protein overexpression and intense EGFR immunostaining in NPC samples with advanced tumor node metastasis stage, clinical stage, and distant metastasis in NPC patients. Exogenous EGF stimulates NPC mobility and invasiveness in vitro. Activation of EGFR signaling prompted PKM2 translocation to the nucleus. Silencing either EGFR or PKM2 attenuates NPC cell aggressiveness in vitro and in vivo. Blocking EGFR signaling with cetuximab suppressed NPC cell invasiveness in vitro and metastatic potential in vivo. Comprehensive analyses of transcriptome profiles indicated that the EGFR-PKM2 axis activates a number of novel metastasis promoters, including F3, FOSL1, EPHA2, ANTXR2, and AKR1C2. Finally, we found that the metastasis-promoting function of the EGFR-PKM2 axis is dependent on nuclear PKM2 regulation of the transcription of metastasis-related genes, including FOSL1 and ANTXR2. Our study indicates that EGFR-PKM2 signaling promotes NPC cell invasion and metastasis through induction of FOSL1 and ANTXR2 and identifies EGFR as a promising biomarker for predicting the risk of distant metastasis.

This study revealed activation of EGFR signaling promotes NPC metastasis via nuclear PKM2 mediated induction of FOSL1 and ANTXR2.

Introduction

Nasopharyngeal carcinoma (NPC) is a unique subtype of head and neck squamous cell carcinoma (HNSCC) prone to distant metastasis and prevalent in South China and South East Asia (1). NPC exhibits a distinct etiological and geographical distribution and commonly metastasizes to the lungs, liver, and bones. The incidence rate of distant metastatic NPC at presentation varies from 7.7 to 20.3% (2–6). Although significant improvement in local control has been achieved by the use of high precision radiotherapy, treatment failure for distant metastasis is still the primary cause of death from NPC. Given the high mortality rate of distant metastatic NPC, it is critical to identify novel targets and develop new strategies for antimetastasis therapy (7–10).

The epidermal growth factor receptor (EGFR) is a transmembrane protein in the ErbB family, a subfamily of four closely related receptor tyrosine kinases. Upon binding with its cognate ligand EGF, intracellular EGFR tyrosine phosphorylation and receptor dimerization with other family members lead to enhanced cell proliferation (11). The EGFR protein is frequently overexpressed in a wide variety of human cancers and promotes cell growth and proliferation. Blocking EGFR signaling by antibodies or inhibition of its intracellular tyrosine kinase activity suppresses tumor cell growth and improves patient condition. Genomic sequencing studies have revealed that EGFR mutations are rare in NPC samples (12–15). Although its growth promotive role is well established, the contribution of EGFR to NPC metastasis remains elusive. Activation of EGFR signaling promotes the invasiveness of NPC cells (16,17); however, the downstream effectors involved in EGFR signaling-mediated tumor invasion and metastasis have not been identified.

Increased dependence on aerobic glycolysis and overexpression of glycolytic enzymes is an emerging hallmark of cancer; this phenomenon is called the Warburg effect (18,19). Pyruvate kinase M2 (PKM2) is the rate-limiting enzyme catalyzing the formation of pyruvate from the conversion of phosphoenolpyruvate. PKM2 is the fetal isoform of pyruvate kinase but is overexpressed in various human cancers (20). PKM2 localizes to the cell nucleus and regulates gene transcription (21). Activation of EGFR signaling induces nuclear translocation of PKM2 and stimulates cyclin D1 (CCND1) transcription (22). Recent evidence suggests a crucial role for PKM2 in pancreatic cancer metastasis (23). Nevertheless, the role of PKM2 in NPC metastasis remains elusive.

The objective of this study was to investigate the association between EGFR-PKM2 signaling and NPC metastasis and the underlying mechanism of action, in order to identify novel targets for antimetastasis therapy for NPC.

Materials and methods

NPC samples and immunohistochemistry

A cohort of clinical samples, including 309 NPC cases and 92 samples of noncancerous inflammatory nasopharyngeal epithelial tissues, was obtained between January 2012 and October 2017 from the Pathology Department of Cancer Hospital affiliated with the Xiangya Medical School, Central South University (Hunan, China). All patients provided signed consent to participate. A polyclonal anti-EGFR antibody was obtained from Maxin, Inc. (Fuzhou, China). EGFR protein was detected by immunohistochemical staining according to methods described previously (8,24). A staining index (values, 0–6) was calculated from the staining intensity (scores: negative = 0, weak = 1, moderate = 2, or strong = 3) and the percentage of stained tumor cells (scores: <10% = 1, 10–50% = 2, >50% = 3). The sum of these two scores was used as the final immunoreactive score (0–6), i.e. low expression (0–2 scores) and high expression (3–6 scores). This study, which involved the use of clinical samples, was approved by the Institute Research Ethics Committee.

Cell lines and culture

A well-differentiated NPC cell line (HK1) (25) and a hypopharyngeal carcinoma cell line (FaDu) (26,27) were routinely maintained in our laboratory. Cells were grown in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) and penicillin/streptomycin (Gibco, Grand Island, NY, USA) in a humidified incubator at 37°C with 5% CO₂ and 95% air. The cells were authenticated by short tandem repeat analysis by Life Technologies every 6 months. Cells were treated with recombinant EGF (SinoBiological, Inc., Beijing, China) at 100 ng/ml or micheliolide (MCL, MedChemExpress, Beijing, China) at 5 μM or cetuximab (Erbitux^®, Merck KGaA, Darmstadt, Germany) at 20 ng/ml.

siRNA, shRNA, and gene transfection

All gene targeting siRNAs and scrambled siRNAs used in this study were purchased from GenePharma (Shanghai, China). By using Lipofectamine® RNAiMAX Reagent (Invitrogen, Carlsbad, CA, USA), siRNAs were transfected into NPC cells according to the manufacturer’s protocol. The siRNA sequences are listed in Supplementary Table S1, available at Carcinogenesis Online.

RNA isolation and real-time reverse transcription PCR (RT-qPCR)

TRIzol Reagent (Invitrogen, San Diego, CA, USA) was used to extract total RNA as previously described (28). Residual genomic DNA in total RNA samples was removed by RNase-free DNase I (Roche Diagnostics, Rotkreuz, Switzerland), and 1 μg of total RNA was reverse transcribed to cDNA using M-MLV reverse transcriptase (Invitrogen, San Diego, CA, USA). The mRNA levels were evaluated by the CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) using SYBR Green I (Selleck, Shanghai, China). The primer sequences are listed in Supplementary Table S2, available at Carcinogenesis Online.

Immunofluorescence and fluorescence-activated cell sorting (FACS) assays

For the immunofluorescence assay, tumor cell suspensions were fixed with 4% paraformaldehyde. After fixation, cetuximab or human IgG was incubated with tumor cells for 24 h at 4°C, and then the cells were stained by FITC-conjugated secondary antibody. The frequency of antibody-labeled tumor cells was measured by flow cytometry (BD Biosciences, San Jose, CA, USA).

RNA-seq

Total RNA was isolated using the TRIzol reagent (Invitrogen). A Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA, USA) was used to monitor the integrity of the RNA samples. RNA-seq was performed with the Illumina-PE150 sequencer (Illumina Inc., San Diego, CA, USA). The NOISeq method was employed to identify differentially expressed genes with a 1.5-fold change between two groups (29). Gene set enrichment analysis (GSEA) was performed as described previously (30).

Chromatin immunoprecipitation (ChIP) and qPCR

For ChIP analysis, cells grown on a 10-cm plate were processed as described in the ChIP Assay kit (17-295) protocol (Millipore, Billerica, MA, USA). The PKM2-associated chromatin was immunoprecipitated by an anti-PKM2 antibody (Cell Signaling Technology; Cat No. 8685S). The precipitated DNA fragments were purified and measured by qPCR under the conditions described above. Primers specific to each segment of interest are listed in Supplementary Table S3, available at Carcinogenesis Online.

Protein extraction and western blotting

Cellular proteins were extracted using a lysis buffer (Beyotime, Jiangsu, China), separated by SDS-PAGE, and then transferred onto PVDF membranes (Millipore). Protein levels were detected with the following antibodies: polyclonal anti-EGFR (Proteintech; Cat No. 18986-1-AP), polyclonal anti-PKM2 (Cell Signaling Technology; Cat No. 4053T), anti-FOSL1 (Abclonal; Cat No. A5372), anti-ANTXR2 (Abclonal; Cat No. A6526), anti-histone H3 (Cell Signaling Technology; Cat No. 4499), and anti-GAPDH (BBI Life Sciences; Cat No. D190090-0100).

Cell migration and invasion assays

Tumor cell mobility or invasiveness was evaluated using 8-μm-pore Transwell inserts (Corning-Costar, Cambridge, MA, USA) precoated with or without 15 μl of Matrigel (BD Biosciences, Bedford, MA, USA), respectively. Briefly, single-cell suspensions in serum-free RPMI-1640 medium (Gibco) were plated onto the upper inserts at a density of 100 000 cells/well and then the inserts were placed into the lower chamber with 600 μl of culture medium containing 15% FBS. Cells in the Transwell inserts were incubated for 6–24 h at 37°C to allow cells to migrate or invade across the membrane. Cells that had migrated or invaded were then visualized with crystal violet dye. The average number of migrated or invaded cells from five random fields was counted.

In vivo tumor metastasis and in vivo bioluminescent assays

All animal experiments were performed according to the guidelines of the Ethical Review Committee of Central South University of China. HK1 cells were infected with a luciferase-expressing lentivirus. Single-tumor cell suspensions (1 × 10⁶/0.2 ml) in serum-free RPMI-1640 were orthotopically transplanted into the livers of 6-week-old male BALB/c nude mice (31,32) (Shanghai SLAC Laboratory Animal Co. Ltd., Shanghai, China). Two weeks posttransplantation, the mice were intraperitoneally injected with cetuximab (Merck Millipore, Darmstadt, Germany) at a dose of 5 mg/kg. hIgG (5 mg/kg) was used as a negative control. The average body weight for each group was determined weekly. After 8 weeks of treatment, the mice were intraperitoneally injected with a luciferin solution at a dose of 150 mg/kg (15 mg/ml in PBS). Metastatic tumors were evaluated once by the in vivo bioluminescent assay at 8 weeks posttransplantation, after which the mice were sacrificed (31); the liver tissues were obtained, fixed in 4% saline-buffered formalin, embedded in paraffin, sectioned at 5 μm and stained with hematoxylin and eosin (H&E).

Statistical analysis

Differences in the quantitative variables between groups were analyzed by Student’s t test. The Pearson’s Chi-square test was used to analyze the association of EGFR expression with clinicopathological characteristics using the SPSS 13.0 software package (SPSS, Chicago, IL, USA). A value of P < 0.05 was considered statistically significant.

Results

Overexpression of EGFR in NPC is associated with advanced clinical stage and distant metastasis

The levels of EGFR protein in NPC samples were evaluated by immunohistochemical staining. As shown in Figure 1A, immunostaining of EGFR protein is restricted in the basal cells in noncancerous nasopharyngeal epithelium, but not in the differentiated pseudostratified multiciliated columnar epithelial cells. Only 17 out of 309 NPC samples (5.5%) exhibited negative staining for EGFR protein (Figure 1B). Moderate (Figure 1C) to high levels (Figure 1D) of membrane staining for EGFR protein are observed in the majority of NPC samples (292 out of 309, 94.5%). Among the EGFR-positive NPC samples, 90 out of 309 (30.42%) exhibit intense membrane staining for EGFR protein. The association of EGFR expression with clinicopathological features was analyzed (Table 1). EGFR expression was not associated with gender, but intense EGFR staining was associated with patient age, advanced tumor node metastasis stage, clinical stage, and distant metastasis status. We also analyzed the association of expression levels of EGFR mRNA with overall survival (OS) in HNSCC. According to The Cancer Genome Atlas (TCGA) dataset, high expression of EGFR mRNA predicts a worse prognosis than low expression of EGFR mRNA (P < 0.01, Figure 1E and F).

Figure 1.

High expression of EGFR is associated with distant metastasis in NPC (A) Immunostaining of EGFR protein is restricted in the basal layer of normal nasopharyngeal epithelium, but not in the differentiated columnar epithelial cells. (B) Negative staining of EGFR in a subset of NPC samples. (C) Moderate immunostaining of EGFR protein in NPC samples without distant metastasis. (D) Intense immunostaining of EGFR protein in NPC samples with distant metastasis. (E) mRNA levels of EGFR are elevated in head and neck cancer. (F) High expression of EGFR predicts unfavorable clinical outcomes in head and neck cancer patients. *p < 0.05).

Table 1.

High expression of EGFR protein is associated with advanced NPC progression and distant metastasis.

Factor	Class	Sample No.	Expression of EGFR		Pearson’s Chi-square test
			Low (0–4)	High (5,6)	χ 2	df	P
Gender	Male	228	156 (68.4%)	72 (31.6%)	0.5513	1	0.4578
	Female	81	59 (72.8%)	22 (27.2%)
Age	＜50	83	65 (78.3%)	18 (21.7%)	4.09	1	0.043
	≥50	226	150 (66.4%)	76 (33.6%)
TNM	T1-T2	138	106 (76.8%)	32 (23.2%)	6.162	1	0.013
	T3-T4	171	109 (63.7%)	62 (36.3%)
Clinical stage	I–III	171	129 (75.4%)	42 (24.6%)	6.210	1	0.0127
	IV	138	86 (62.3%)	52 (37.7%)
Metastasis	No	257	187 (72.8%)	70 (27.2%)	7.312	1	0.0069
	Yes	52	28 (53.9%)	24 (46.1%)

Activation of EGFR signaling promotes migration and invasiveness of NPC in vitro

The association of EGFR levels with distant metastasis prompted us to ask whether overexpression of EGFR contributes to higher mobility and invasiveness in NPC. As shown in Figure 2A and B, recombinant EGF treatment stimulates migration and invasiveness of HK1 and FaDu cells. Transient transfection of two separate siRNAs targeting EGFR successfully inhibits the expression of EGFR in HK1 and FaDu cells at both the mRNA and protein level (Figure 2C). As a result, transient silencing of EGFR expression by two siRNAs markedly inhibits the migration and invasiveness of HK1 and FaDu cells (Figure 2D and E). When EGFR mRNA and protein levels in HK1 and FaDu cells were stably knocked down by shRNA-expressing lentivirus (Figure 2F), the migration and invasion of NPC cells were markedly inhibited (Figure 2G and H).

Figure 2.

Activation of EGFR signaling promotes NPC cell migration and invasiveness in vitro. (A,B) Exogenous EGF (100 ng/ml) treatment stimulates migration (A) and invasiveness (B) of NPC cells. (C) EGFR was transiently silenced in NPC cells, and mRNA and protein levels of EGFR were measured by RT-PCR and western blot analysis, respectively. (D,E) Transient silencing of EGFR inhibits migration (D) and invasiveness (E) of NPC cells in vitro. (F) EGFR was stably silenced in NPC cells, and mRNA and protein levels of EGFR were measured by RT-PCR and western blot analysis, respectively. (G, H) Stable silencing of EGFR inhibits migration (G) and invasiveness (H) of NPC cells in vitro. * P < 0.05. ** P < 0.01. *** P < 0.001.

Blockage of EGFR signaling by cetuximab inhibits metastasis in vivo

We then investigated the effects of blocking EGFR signaling by the EGFR monoclonal antibody cetuximab on NPC cell migration and invasion and found that cetuximab prevents invasiveness in vitro (Figure 3A and B). We also measured the binding of cetuximab on the tumor cell surface by FACS assays. The results showed that more than 99% of HK1 cells, compared to 76.5% of FaDu cells, bind cetuximab (Supplementary Figure S1, available at Carcinogenesis Online), which is in consistent with the results showing that HK1 cells are more sensitive to cetuximab-mediated suppression of migration and invasiveness than FaDu cells (Figure 3A and B). Hepatic subcapsular transplantation was employed to evaluate whether blockage of EGFR signaling by cetuximab inhibits NPC cells metastasis in vivo. HK1 cells infected with a luciferase-expressing lentivirus were transplanted into the livers of nude mice. Two weeks after transplantation, cetuximab was intraperitoneally injected at 1 mg/kg weekly. hIgG was used as a negative control. During the cetuximab treatment period, we observed no difference in the weight of the mice between groups (Figure 3C). After 8 weeks of treatment, hepatic metastasis evaluated by an in vivo bioluminescent assay revealed that cetuximab treatment effectively suppressed HK1 cell metastasis of the livers of nude mice (Figure 3D). Morphological analysis (Figure 3E) and H&E staining (Figure 3F) confirmed that there were fewer metastatic tumor loci in the livers of cetuximab-treated mice than in the hIgG-treated control group.

Figure 3.

Blocking EGFR signaling by cetuximab suppresses NPC cell invasiveness in vitro and metastasis in vivo. (A, B) Cetuximab suppresses migration (A) and invasiveness (B) of NPC cells in vitro. (C) Luciferase-expressing HK1 cells were transplanted into the livers of nude mice and then the cetuximab-treated and untreated (control) mice were weighed. (D) Noninvasive bioluminescence imaging of luciferase-expressing intrahepatic HK1 xenografts showed that cetuximab suppresses orthotopic hepatic metastatic tumor formation. (E) A representative macroscopic image of control and cetuximab-treated nude mice. The tumor nodes were counted, and the data showed that cetuximab treatment reduced the number of liver tumor nodes. (F) H&E staining showed decreased metastatic tumor formation in the liver of cetuximab-treated mice. * p < 0.05. ** p < 0.01.

Nuclear PKM2 contributes to EGFR signaling-mediated NPC cell migration and invasion

It has been demonstrated that PKM2 translocates to the nucleus and functions as a transcription cofactor upon activation of EGFR signaling (22,33). To study the role of PKM2 in NPC cell migration and invasion, the mRNA and protein levels of PKM2 in HK1 and FaDu cells were reduced by two separate siRNAs transfections (Supplementary Figure S2A, available at Carcinogenesis Online). Transient silencing of PKM2 in HK1 and FaDu cells resulted in a significant decrease in mobility and invasiveness in vitro (Supplementary Figure S2B and C, available at Carcinogenesis Online). Then, the mRNA and protein levels of PKM2 in HK1 and FaDu cells were stably inhibited by shRNA-expressing lentivirus (Supplementary Figure S2D, available at Carcinogenesis Online). Stable depletion of PKM2 dramatically attenuated tumor cell migration and invasion in vitro (Supplementary Figure S2E and F, available at Carcinogenesis Online), suggesting that the expression of PKM2 in NPC cells contributes to increased cell mobility and invasiveness. We then investigated whether PKM2 acts downstream of EGFR signaling in NPC cells. Western blot analysis revealed that silencing of EGFR by two siRNAs leads to decreased nuclear PKM2 protein levels in HK1 and FaDu cells (Figure 4A), whereas EGF treatment promotes nuclear PKM2 expression (Figure 4B). We then demonstrated that EGF-stimulated NPC cell migration and invasiveness are dramatically suppressed by the loss of PKM2 (Figure 4C and D). Micheliolide (MCL) is a natural product that binds specifically to PKM2 and prevents its nuclear translocation (34). The western blot assay revealed that MCL treatment inhibited nuclear translocation of PKM2 protein in HK1 cells (Supplementary Figure S3, available at Carcinogenesis Online). More importantly, MCL treatment suppresses EGFR-stimulated NPC cell migration and invasiveness in vitro (Figure 4E and F), suggesting that nuclear PKM2 contributes significantly to EGFR signaling-mediated NPC invasion. Furthermore, we demonstrated that high expression of nuclear PKM2 protein is associated with distant metastasis in NPC patients (Supplementary Table S4, available at Carcinogenesis Online); however, the association of total PKM2 protein level to distant metastasis was not significant (Supplementary Table S5, available at Carcinogenesis Online). Both total PKM2 and nuclear PKM2 protein level were associated with unfavorable OS and disease-free survival (DFS) in NPC patients (Supplementary Figure S4, available at Carcinogenesis Online).

Figure 4.

Activation of EGFR signaling promotes PKM2 nuclear translocation in NPC cells. (A) Transient silencing of EGFR leads to reduced nuclear PKM2 protein in HK1 and FaDu cells. The immune-intensity levels of cytoplasmic or nuclear PKM2 protein were normalized to GAPDH or histone 3 (H3), respectively. (B) Exogenous EGF treatment (100 ng/ml) promotes PKM2 nuclear translocation. (C, D) Depletion of PKM2 impairs EGF stimulated NPC cell migration (C) and invasion (D). (E, F) Inhibition of PKM2 nuclear translocation by micheliolide (MCL) treatment suppresses EGF-stimulated NPC cell migration (E) and invasion (F). * p < 0.05. ** p < 0.01. *** p < 0.001.

Depletion of either EGFR or PKM2 suppresses NPC cell metastasis in vivo

Next, we examined whether activation of EGFR-PKM2 signaling was required for tumor metastasis in vivo. A subhepatic transplantation model was employed to evaluate the metastatic potential of the EGFR- or PKM2-depleted HK1 cells. Either EGFR- or PKM2-depleted HK1 cells were orthotopically transplanted into the livers of nude mice. The weights of mice were recorded, and the data showed no difference between the three groups (Figure 5A and B). Eight weeks posttransplantation, the mice were sacrificed and liver tissues were obtained, fixed in 4% saline-buffered formalin, embedded in paraffin, sectioned at 5 μm, and stained with H&E. Morphological analysis (Figure 5C) and H&E staining (Figure 5D) confirmed that stable depletion of either EGFR or PKM2 effectively reduces liver metastatic tumor formation in the nude mice.

Figure 5.

Stable silencing of either EGFR or PKM2 suppresses NPC cell metastasis in vivo. (A) A macroscopic image of nude mice transplanted with control cells, EGFR-depleted, or PKM2-depleted HK1 cells. (B) Body weights of mice transplanted with control cells, EGFR-depleted, or PKM2-depleted HK1 cells. (C) A macroscopic image of livers from nude mice transplanted with control cells, EGFR-depleted, or PKM2-depleted HK1 cells; arrows indicate metastatic tumor foci. (D) H&E staining shows metastatic NPC cells in the livers of mice transplanted with control cells and decreased metastatic tumor formation in the livers of mice transplanted with either EGFR-depleted or PKM2-depleted HK1 cells.

Transcriptomic alterations induced by loss of EGFR-PKM2 signaling

We employed RNA-seq to search for the downstream effectors of EGFR-PKM2 signaling in NPC cells. A total of 648 genes are consistently upregulated, and 630 genes consistently downregulated, following loss of either EGFR or PKM2 expression in HK1 cells (Supplementary Figure S5A, available at Carcinogenesis Online). Among these downregulated genes, CCND1 was previously identified as a major target of EGFR-PKM2 signaling (22,33), indicating the robustness of our RNA profiling approach. Bioinformatics analysis (Supplementary Figure S5B and C, available at Carcinogenesis Online) revealed that the upregulated genes are mainly involved in the Hippo, Foxo, and melanogenesis signaling pathways regulating pluripotency. The downregulated genes are associated with cytokine–cytokine receptor interaction, proteoglycans in cancer, MAPK signaling, Hippo signaling, and focal adhesion. GSEA (Supplementary Figure S5D, available at Carcinogenesis Online) further revealed that the downregulated genes are positively associated with Wnt/β-catenin signaling and epithelial–mesenchymal transition processes, which are involved in cancer metastasis (35–37). A number of differentially expressed genes were selected for further validation by RT-PCR (Supplementary Figure S5E, available at Carcinogenesis Online). For example, the mRNA levels of FOSL1, ANTXR2, CCND1, F3, EPHA2, and Vimentin are consistently decreased following EGFR or PKM2 depletion in HK1 and FaDu cells.

FOSL1 and ANTXR2 contribute to EGF-stimulated NPC cell migration and invasion

Among those differentially expressed gene coregulated by EGFR and PKM2, FOSL1, and ANTXR2 are recently suspected to promote cancer progression (38,39). Western blot analysis showed that FOSL1 and ANTXR2 protein levels are downregulated in EGFR- or PKM2-silenced NPC cells (Figure 6A). To assess the biological function of FOSL1 and ANTXR2 in NPC invasion, either FOSL1 or ANTXR2 was inhibited by specific siRNAs. Depletion of either FOSL1 or ANTXR2 significantly suppresses EGF-stimulated NPC HK1 cell migration and invasion in vitro (Figure 6B), suggesting they are bonafide downstream effectors of EGF signaling-mediated NPC cell migration and invasion. Interestingly, both FOSL1 and ANTXR2 mRNAs levels are overexpressed in head and neck cancer samples according to the TCGA data. ChIP-qPCR assays revealed that PKM2 binds to the promoter regions of FOSL1 and ANTXR2, whereas loss of EGFR led to a decrease of PKM2 binding to those regions (Figure 6C). Similarly, silencing EGFR reduces PKM2 binding to the promoter of CCND1, which was previously identified as a target of the EGFR-PKM2 axis (33), confirming the robustness of our data. Transient silencing of PKM2 significantly attenuated EGF treatment induced FOSL1 and ANTXR2 mRNA levels in HK1 and FaDu cells (Figure 6D), indicating that nuclear PKM2 at least partially contributes to transactivation of FOSL1 and ANTXR2 upon EGFR activation. By analyzing TCGA data, we demonstrated a positive association between the mRNA levels of either FOSL1 or ANTXR2 with EGFR or PKM2 in head and neck cancer samples (Figure 6E). We also measured these protein levels in a subset of NPC samples by immunohistochemical staining. The results clearly demonstrated that the protein levels of FOSL1 and ANTXR2 were positively correlated with EGFR and PKM2 (Supplementary Figure S6). More importantly, high expression of FOSL1 or ANTXR2 predicts an unfavorable prognosis in head and neck cancer patients (Figure 6F). Thus, our data strongly suggested that FOSL1 and ANTXR2 act as downstream effectors of EGFR-PKM2 signaling in NPC metastasis.

Figure 6.

FOSL1 and ANTXR2 contribute to EGFR-PKM2 signaling-mediated tumor invasiveness. (A) Western blot analysis revealed a decrease in FOSL1 and ANTXR2 protein levels after silencing of EGFR and PKM2. (B) Silencing of either FOSL1 or ANTXR2 dramatically represses EGF-stimulated HK1 tumor cell migration and invasion. (C) Silencing of EGFR inhibited PKM2 binding to the promoter regions of FOSL1 and ANTXR2; CCND1 was used as a positive control. (D) Silencing PKM2 attenuated EGF-induced FOSL1 and ANTXR2 mRNA expression in HK1 and FaDu cells. (E) mRNA levels of FOSL1 and ANTXR2 are positively correlated with either EGFR or PKM2 mRNA levels. (F) High expression of FOSL1 or ANTXR2 predicts an unfavorable prognosis in head and neck cancer patients. * p < 0.05. ** p < 0.01. *** p < 0.001.

Discussion

Distant metastasis occurs in approximately 20–30% of NPC patients (40). The prognosis of NPC patients with distant metastasis is very poor due to a lack of effective therapies (41). ¹⁸F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) is the most sensitive method for detecting distant metastasis of NPC, especially for bone metastasis. However, the efficacy of PET/CT in detecting liver metastasis is still suboptimal (42). For patients at high risk of metastasis, such as those with high nodal stage and nodal volume, assessment of the risk of distant metastasis before initial treatment of NPC is still a great challenge (43).

In this study, we found that intense EGFR immunostaining is associated with an increased risk of distant metastasis in NPC. We also provide evidence that blocking EGFR signaling efficiently suppressed NPC cell invasiveness in vitro and metastasis in vivo. Our data strongly suggest that EGFR is a promising biomarker for predicting the risk of distant metastasis. We envision the development of an EGFR immunostaining score model for assessing the risk of distant metastasis. For patients with extremely high levels of EGFR, in whom the metastasis cannot be imaged, more active systemic treatments or preventive treatments should be considered. A recent meta-analysis demonstrated that anti-EGFR monoclonal antibody treatment combined with radiotherapy or chemoradiotherapy significantly improved OS and DFS in NPC patients (44) with tolerable side effects. This suggests that locoregionally advanced NPC would likely benefit from anti-EGFR monoclonal antibody treatment. Our preclinical experiment showed that cetuximab effectively suppresses NPC cell metastasis in nude mice. Given that limited therapeutic options are available for NPC patients with distant metastasis after chemotherapy failure, our study supports the current treatment paradigm that blockage of EGFR signaling by anti-EGFR monoclonal antibodies is a feasible treatment for metastatic NPC.

Importantly, our preliminary studies demonstrate that nuclear PKM2 contributes significantly to increased mobility and invasiveness of NPC cells upon EGFR activation. Either total PKM2 or nuclear PKM2 protein level predicts unfavorable prognosis of NPC patients. We also provide evidence that MCL treatment inhibits NPC cell migration and invasion. MCL is a natural compound that binds with and selectively activates PKM2 (34). It has been shown that while PKM2 tetramers are distributed in the cytoplasm, PKM2 dimers are localized to the nucleus where they act as active protein kinases and regulate gene transcription and stimulate cancer cell proliferation (21). Covalent binding at cysteine residue 424 (C424) of PKM2 with MCL promotes its tetramer formation, thus inhibiting the translocation of PKM2 into the nucleus. An MCL derivate, dimethylaminomicheliolide, exerts growth-inhibitory effects on leukemia and glioma cells and is currently in clinical trial in Australia (34,45). We believe it is worth evaluating the antimetastatic potential of MCL in EGFR^-overexpressing NPC cells.

By using a transcriptomic approach, a number of downstream target genes of EGFR and PKM2 were identified, and we selected FOSL1 and ANTXR2 for further functional validation. Silencing either FOSL1 or ANTXR2 suppressed EGF-stimulated NPC cell migration and invasion. It has been demonstrated that PKM2 regulates gene transcription through direct binding to the promoters of target genes, for example, MEK5 (21). The ability of PKM2 to regulate transcription is mainly dependent on PKM2-mediated protein phosphorylation. Upon EGFR signaling activation, PKM2 binds to phosphorylated β-catenin, facilitating the recruitment of both proteins to the CCND1 promoter (22). Nuclear PKM2 phosphorylates histone H3 at T11 and promotes acetylation of histone H3 at lysine 9, leading to activation of CCND1 and c-myc transcription (33).

In this study, we observed that PKM2 binds to the promoter regions of FOSL1 and ANTXR2, and blocking EGFR signaling reduces PKM2 binding to these regions. Along with FOSL1 and ANTXR2, depletion of EGFR also reduces PKM2 binding to the CCND1 promoter. It is not clear which transcription factor cooperates with PKM2 in the process of EGF-induced FOSL1 and ANTXR2 expression. FOSL1 is a subunit of the AP-1 transcription factor complex. It was recently revealed that FOSL1 is required for KRAS-driven lung and pancreatic cancer. FOSL1 activates Aurora kinase A (AURKA) transcription, linking the KRAS oncogene to components of the mitotic machinery (46). FOSL1 also acts as an oncogene and transforms melanocytes through enhancement of several growth-promoting processes (47). Consistent with these observations, our findings suggest that FOSL1 is an attractive therapeutic target for various cancers, not only for controlling tumor growth but also for treatment of metastasis. ANTXR2 is also known as capillary morphogenesis protein 2 (CMG2) and participates in capillary morphogenesis. Overexpression of ANTXR2 is associated with poor prognosis of glioma patients. ANTXR2 promotes migration and invasion of glioma cells partially through enhancement of Yes-associated protein 1 (YAP1) activity (48). High expression of ANTXR2 is also associated with the invasion depth and lymph node metastasis of gastric cancer. Depletion of ANTXR2 in gastric cancer stem-like cells attenuates their self-renewing, invasive, and metastatic capabilities (39). In this study, we demonstrated that both FOSL1 and ANTXR2 are overexpressed in head and neck cancer, and their expression levels are positively correlated with either EGFR or PKM2 levels and predict unfavorable clinical outcomes of patients. The roles and underlying mechanisms of action of FOSL1 and ANTXR2 in NPC metastasis require further investigation.

In conclusion, we provide evidence that overexpression of EGFR is associated with distant metastasis in NPC. Upon EGFR signaling activation, PKM2 translocates to the nucleus and activates FOSL1 and ANTXR2 transcription, leading to increased migration and invasiveness in NPC. Our findings may contribute to the development of novel antimetastasis therapy for advanced NPC.

Funding

This work was supported in part by grants from The National Natural Science Foundation of China (81572667, 81772902, 81872278, 81703131); the National “111” Project (Project #111–2–12); The Natural Science Foundation of Hunan Province, China (2018JJ1040); and Hunan Provincial Key Research and Development Program (2018SK2130, 2018SK2131).

Conflict of interest statement: None declared.

Glossary

Abbreviations

AA

arachidonic acid

EGFR

epidermal growth factor receptor

FBS

fetal bovine serum

GSEA

gene set enrichment analysis

NPC

Nasopharyngeal carcinoma

OS

overall survival

PKM2

pyruvate kinase M2