Therapeutic targeting of the USP2-E2F4 axis inhibits autophagic machinery essential for zinc homeostasis in cancer progression

Abstract

Macroautophagy/autophagy is a conserved cellular process associated with tumorigenesis and aggressiveness, while mechanisms regulating expression of autophagic machinery genes in cancers still remain elusive. Herein, we identified E2F4 (E2F transcription factor 4) as a novel transcriptional activator of cytoprotective autophagy crucial for zinc homeostasis in cancer cells. Gain- and loss-of-function studies showed that E2F4 promoted autophagy in a cell cycle-dependent manner, resulting in facilitated degradation of MT (metallothionein) proteins, elevated distribution of Zn²⁺ within autophagosomes, decreased labile intracellular zinc ions, and increased growth, invasion, and metastasis of gastric cancer cells. Mechanistically, E2F4 directly regulated the transcription of ATG2A (autophagy related 2A) and ULK2 (unc-51 like autophagy activating kinase 2), leading to autophagic degradation of MT1E, MT1M, and MT1X, while USP2 (ubiquitin specific peptidase 2) stabilized E2F4 protein to induce its transactivation via physical interaction and deubiquitination in cancer cells. Rescue experiments revealed that USP2 harbored oncogenic properties via E2F4-facilitated autophagy and zinc homeostasis. Emetine, a small chemical inhibitor of autophagy, was able to block interaction between UPS2 and E2F4, increase labile intracellular zinc ions, and suppress tumorigenesis and aggressiveness. In clinical gastric cancer specimens, both USP2 and E2F4 were upregulated and associated with poor outcome of patients. These findings indicate that therapeutic targeting of the USP2-E2F4 axis inhibits autophagic machinery essential for zinc homeostasis in cancer progression.Abbreviations: 3-MA: 3-methyladenine; ANOVA: analysis of variance; ATG2A: autophagy related 2A; ATG5: autophagy related 5; ATP: adenosine triphosphate; BECN1: beclin 1; BiFC: bimolecular fluorescence complementation; CCND1: cyclin D1; CDK: cyclin dependent kinase; ChIP: chromatin immunoprecipitation; CHX: cycloheximide; Co-IP: co-immunoprecipitation; DAPI: 4',6-diamidino-2-phenylindole; E2F4: E2F transcription factor 4; eATP: extracellular adenosine triphosphate; EBSS: Earle's balanced salt solution; FP: first progression; FRET: fluorescence resonance energy transfer; FUCCI: fluorescent ubiquitination-based cell cycle indicator; GFP: green fluorescent protein; GST: glutathione S-transferase; HA: hemagglutinin; MAP1LC3B/LC3B: microtubule associated protein 1 light chain 3 beta; MDM2: MDM2 proto-oncogene; MKI67/Ki-67: marker of proliferation Ki-67; MT: metallothionein; MT1E: metallothionein 1E; MT1M: metallothionein 1M; MT1X: metallothionein 1X; MTT: 3-(4,5-dimethyltriazol-2-yl)-2,5-diphenyl tetrazolium bromide; OS: overall survival; PECAM1/CD31: platelet and endothelial cell adhesion molecule 1; PIK3C3: phosphatidylinositol 3-kinase catalytic subunit type 3; qPCR: quantitative PCR; RFP: red fluorescent protein; SQSTM1/p62: sequestosome 1; UBXN1: UBX domain protein 1; Ub: ubiquitin; ULK2: unc-51 like autophagy activating kinase 2; USP14: ubiquitin specific peptidase 14; USP2: ubiquitin specific peptidase 2; USP5: ubiquitin specific peptidase 5; USP7: ubiquitin specific peptidase 7; ZnCl₂: zinc chloride.

Keywords: Autophagy; E2F transcription factor 4; gastric cancer; ubiquitin specific peptidase 2; zinc homeostasis.

Figure 1.

Identification of E2F4 as a transcription factor activating autophagic genes in gastric cancer. (A) Venn diagram (left panel) and table (right panel) showing discovery of macroautophagic/autophagic genes associated with overall and first progression survival of gastric cancer patients derived from KM plotter database (http://kmplot.com), transcription factors (TF) regulating expression of these autophagic machinery genes analyzed by ChIP-X program, and those associated with overall and first progression survival. (B) Kaplan-Meier curves indicating overall (OS) and first progression (FP) survival of gastric cancer cases with low or high levels of E2F4 (cutoff values = 7.02 and 6.61). (C) Relative levels of E2F4 in gastric cancer cases (TCGA) with different pathological stages (t2a vs. t3, t2a vs. t4a). (D) Western blot assay showing the protein levels of E2F4 in normal gastric mucosa and cancer cell lines. (E) Real-time qRT-PCR assay revealing the transcript levels of target genes (normalized to ACTB/β-actin) in MGC-803 and AGS cells stably transfected with empty vector (mock), E2F4, scramble shRNA (sh-Scb), or sh-E2F4 (n = 5). (F) ChIP and qPCR assays showing the enrichment of E2F4 (normalized to input DNA) on promoter region of ATG2A and ULK2 in MGC-803 and AGS cells stably transfected with mock, E2F4, sh-Scb, or sh-E2F4 (n = 5). (G and H) Dual-luciferase assay using reporter with wild-type (WT) or mutant (Mut) E2F4 binding site (G) and Western blot assay (H) indicating the promoter activity and expression of ATG2A and ULK2 in MGC-803 and AGS cells stably transfected with mock, E2F4, sh-Scb, or sh-E2F4 (n = 5). Fisher’s exact test for overlapping analysis in A. Log-rank test for survival comparison in B. Bars were means and whiskers (min to max) in C. Student’s t test and ANOVA compared the difference in C and E-G. * P < 0.05 vs. mock or sh-Scb. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in D-H.

Figure 2.

E2F4 facilitates cytoprotective autophagy via ATG2A or ULK2 in gastric cancer. (A) Western blot assay showing the levels of LC3-I, LC3-II, BECN1/Beclin 1, or SQSTM1/p62 in MGC-803 cells stably transfected with empty vector (mock), E2F4, scramble shRNA (sh-Scb), or sh-E2F4. (B) Representative images (left panel) and quantification (right panel) of GFP-LC3 puncta in MGC-803 cells transfected with mock, E2F4, sh-Scb, or sh-E2F4, with nucleus staining by DAPI. Scale bars: 10 μm. (C) Representative images (upper panel) and quantification (lower panel) of autophagic flux reporter RFP-GFP-LC3 in MGC-803 cells stably transfected with mock, E2F4, sh-Scb, or sh-E2F4, and those treated with rapamycin (Rapa, 1 μmol·L⁻¹) or chloroquine (CQ, 20 μmol·L⁻¹) for 4 h as positive or negative controls. Scale bars: 10 μm. (D) Transmission electron microscopic observation and quantification of autophagosomes or autolysosomes (red arrowheads) in MGC-803 cells stably transfected with mock, E2F4, sh-Scb, or sh-E2F4. Scale bars: 2 μm. (E and F) Representative images (upper panel) and quantification (lower panel) of autophagic flux reporter RFP-GFP-LC3 (E) and Western blot assay of target gene expression (F) in MGC-803 cells stably transfected with mock, E2F4, sh-Scb, or sh-E2F4 #1, and those co-transfected with dCas9i control (dCas9i-CTL), dCas9i-ATG2A, or dCas9i-ULK2. Scale bars: 10 μm. Student’s t test and ANOVA compared the difference in B-E. *P < 0.05, **P< 0.01, ***P< 0.001 vs. mock, sh-Scb, or mock+dCas9i-CTL. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in A-F.

Figure 3.

E2F4 decreases metallothionein and zinc ion levels via inducing autophagy. (A) Venn diagram (left panel) and GO pathway analysis (right panel) showing the overlapping of mass spectrometry results from MGC-803 cells treated with stable transfection of E2F4 or 3-MA (1.0 μmol·L⁻¹). (B) Western blot assay indicating the expression of LC3-I, LC3-II, BECN1/Beclin 1, SQSTM1/p62, MT1E, MT1M, MT1X in MGC-803 and AGS cells stably transfected with empty vector (mock), E2F4, scramble shRNA (sh-Scb), or sh-E2F4, and those treated with 3-MA (1.0 μmol·L⁻¹) or ATP (0.1 mmol·L⁻¹). (C) Confocal images of FluoZin-3 staining within MGC-803 and AGS cells stably transfected with mock, E2F4, sh-Scb, or sh-E2F4, and those treated with 3-MA (1.0 μmol·L⁻¹) or ATP (0.1 mmol·L⁻¹). Scale bars: 10 μm. (D) Representative images (left panel) and quantification (right panel) of RFP-GFP-LC3 and FluoZin-3 reporters in MGC-803 cells stably transfected with mock, E2F4, sh-Scb, or sh-E2F4, and those treated with 3-MA (1.0 μmol·L⁻¹) or ATP (0.1 mmol·L⁻¹). Scale bars: 10 μm. (E) Intracellular levels of zinc ion within MGC-803 and AGS cells stably transfected with mock, E2F4, sh-Scb, or sh-E2F4, and those treated with 3-MA (1.0 μmol·L⁻¹) or ATP (0.1 mmol·L⁻¹). (F) Representative images (left panel) and quantification (right panel) of FRET assay using eCALWY-4 in MGC-803 and AGS cells stably transfected with mock, E2F4, sh-Scb, or sh-E2F4, and those treated with 3-MA (1.0 μmol·L⁻¹) or ATP (0.1 mmol·L⁻¹). Scale bars: 5 μm. ANOVA compared the difference in D-F. * P < 0.05 vs. mock+DMSO or sh-Scb+PBS. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in B-F.

Figure 4.

E2F4 promotes tumorigenesis and aggressiveness in an autophagy-dependent manner. (A and B) Representative images (upper panel) and quantification (lower panel) of soft agar (A) and matrigel invasion (B) assays indicating the anchorage-independent growth and invasion of MGC-803 and AGS cells stably transfected with empty vector (mock), E2F4, scramble shRNA (sh-Scb), or sh-E2F4, and those treated with 3-MA (1.0 μmol·L⁻¹) or ATP (0.1 mmol·L⁻¹; n = 4). (C and D) Representative images (left panel) and quantification (right panel) of soft agar (C) and matrigel invasion (D) assays showing the anchorage-independent growth and invasion of MGC-803 cells stably transfected with mock or E2F4, and those treated with ZnCl₂ (50 μmol·L⁻¹). (E) Representative images (left panels), tumor growth curve (right upper panel), and weight at the end points (right lower panel) of xenografts formed by subcutaneous injection of MGC-803 cells stably transfected with mock, E2F4, sh-Scb, or sh-ATG5 #2 into the dorsal flanks of nude mice (n = 5 for each group). (F) Representative images (upper panel) and quantification (lower panel) of immunohistochemical staining revealing the expression of MKI67/Ki-67 and PECAM1/CD31 (arrowheads) within xenografts formed by subcutaneous injection of MGC-803 cells stably transfected with mock, E2F4, sh-Scb, or sh-ATG5 #2. Scale bars: 100 μm. (G) Real-time qRT-PCR (normalized to ACTB/β-actin) assay revealing the levels of E2F4, autophagic target genes (ATG2A and ULK2), and metallothioneins (MT1E, MT1M, and MT1X) in subcutaneous xenografts formed by MGC-803 cells stably transfected with mock, E2F4, sh-Scb, or sh-ATG5 #2 (n = 5). (H) Relative zinc ion levels within xenografts formed by subcutaneous injection of MGC-803 cells stably transfected with mock, E2F4, sh-Scb, or sh-ATG5 #2 (n = 5). (I) Representative images (upper left panel), hematoxylin-eosin (HE) staining (upper right panel), quantification of lung metastatic colonization (lower left panel), and survival curve (lower right panel) of nude mice treated with tail vein injection of MGC-803 cells stably transfected with mock, E2F4, sh-Scb, or sh-ATG5 #2 (n = 5 for each group). Student’s t test and ANOVA compared the difference in A-I. Log-rank test for survival comparison in I. * P < 0.05 vs. mock+DMSO, sh-Scb+PBS, mock+PBS, or mock+sh-Scb. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in A-D.

Figure 5.

USP2 stabilizes E2F4 protein via deubiquitination in gastric cancer cells. (A) Coomassie Brilliant Blue staining, co-IP, and mass spectrometry (MS) assays showing proteins immunoprecipitated by E2F4 antibody in MGC-803 cells, and those overlapped with proteins involved in deubiquitination. (B) Co-IP and Western blot assays indicating endogenous interaction between E2F4 and USP2 protein in MGC-803 cells. The IgG-bound proteins were taken as negative controls. (C) Representative images (left panel) and quantification (right panel) of immunofluorescence showing co-localization between E2F4 and USP2 in AGS cells, and those stably transfected with empty vector (mock) or USP2. Scale bars: 10 μm. (D) Co-IP and Western blot assays indicating interaction between E2F4 and USP2 protein in MGC-803 cells transfected with truncations of Flag-tagged USP2 or HA-tagged E2F4 as indicated. (E) Representative images (upper panel) and quantification (lower panel) of BiFC assay showing physical interaction of E2F4 with USP2 in MGC-803 cells co-transfected with VC155-USP2 and VN173-E2F4. Scale bars: 10 μm. (F) Western blot assay revealing the levels of E2F4 in gastric cancer cells stably transfected with mock, USP2, scramble shRNA (sh-Scb), or sh-USP2 #1. (G) Western blot assay indicating the expression of E2F4 in MGC-803 and AGS cells stably transfected with mock, USP2, sh-Scb, or sh-USP2 #1, and those treated with cycloheximide (CHX, 20 μg/ml) or MG132 (10 μmol·L⁻¹) for 4 h. (H) Western blot assay showing the protein levels of USP1 and E2F4 in AGS cells stably transfected with mock or USP2, and those treated with CHX (20 μg/ml) for durations as indicated (n = 5). (I) Co-IP and Western blot assays revealing ubiquitination of E2F4 in MGC-803 and AGS cells transfected with mock, USP2, sh-Scb, sh-USP2 #1, HA-tagged E2F4 with wild-type (WT) or mutant ubiquitination site at lysine 350 (K350), or Flag-tagged Ub. ANOVA compared the difference in C, E and H. *P < 0.05 vs. VN173+ VC155 or mock. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in B-I.

Figure 6.

USP2 harbors oncogenic properties via facilitating E2F4-mediated autophagy. (A) ChIP and qPCR (normalized to input DNA, n = 4), dual-luciferase reporter (n = 5), and real-time qRT-PCR (normalized to ACTB/β-actin, n = 4) assays indicating E2F4 enrichment, promoter activity, and transcript levels of ATG2A and ULK2 in AGS cells stably transfected with empty vector (mock) or USP2, and those co-transfected with scramble shRNA (sh-Scb) or sh-E2F4 #1. (B) Western blot assay showing the protein levels of ATG2A and ULK2 in AGS cells stably transfected with mock or USP2, and those co-transfected with sh-Scb or sh-E2F4 #1. (C) Western blot assay revealing protein levels of LC3-I, LC3-II, BECN1/Beclin 1, and SQSTM1/p62 in AGS cells stably transfected with mock or USP2, and those co-transfected with sh-Scb or sh-E2F4 #1. (D) Representative images of RFP-GFP-LC3 fluorescence punctate in MGC-803 cells stably transfected with mock or USP2, and those co-transfected with sh-Scb or sh-E2F4 #1. Scale bars: 10 μm. (E) Western blot assay indicating the levels of USP2, E2F4, ATG2A, and ULK2 in HEK293 cells with wild-type (WT) or mutant knockout (KO) of E2F4, and those transfected with mock or USP2. (F and G) Representative images (left panel) and quantification (right panel) of soft agar (F) and matrigel invasion (G) assays indicating Anchorage-independent growth and invasion capability of AGS cells stably transfected with mock or USP2, and those co-transfected with sh-Scb or sh-E2F4 #1 (n = 5). (H and I) Representative images (H, left panels), tumor growth curve (H, right upper panel), weight at the end points (H, right lower panel), relative zinc ion levels (H, right lower panel), and immunohistochemical staining of MKI67/Ki-67 and PECAM1/CD31 (I) within xenografts in nude mice formed by subcutaneous injection of AGS cells stably transfected with mock or USP2, and those co-transfected with sh-Scb or sh-E2F4 #1 (n = 5 for each group). Scale bars: 100 μm. (J) Representative images (upper panel) and quantification of lung metastatic colonies (middle panel), and Kaplan-Meier curves (lower panel) of nude mice treated with tail vein injection of AGS cells stably transfected with mock or USP2, and those co-transfected with sh-Scb or sh-E2F4 #1 (n = 5 for each group). ANOVA compared the difference in A and F-J. Log-rank test for survival comparison in J. *P < 0.05 vs. mock+sh-Scb. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in A-G.

Figure 7.

Therapeutic knockdown of USP2 suppresses the tumorigenesis and aggressiveness of gastric cancer. (A and B) Representative images, tumor growth curve, weight at the end points, relative zinc ion levels (A), and immunohistochemical staining of MKI67/Ki-67 and PECAM1/CD31 (B) of xenografts formed by MKN-45 cells in athymic nude mice (n = 5 per group) that received intravenous injection of lentivirus (LV)-mediated scramble shRNA (sh-Scb) or sh-USP2 #1, with body weight (left lower panel) as indicated. Scale bars: 100 μm. (C and D) Western blot (C) and real-time qRT-PCR (D, normalized to ACTB/β-actin) assays revealing the expression of USP2, E2F4, ATG2A, ULK2, MT1E, MT1M, and MT1X within subcutaneous xenografts formed by MKN-45 cells in athymic nude mice (n = 5 per group) that received intravenous injection of LV-mediated sh-Scb or sh-USP2 #1. (E and F) Representative images (E), hematoxylin-eosin (HE) staining (F, left panel), quantification of lung metastatic colonization (F, middle panel), and survival curves (F, right panel) of nude mice (n = 5 for each group) treated with tail vein injection of MKN-45 cells and subsequent administration of LV-mediated sh-Scb or sh-USP2 #1 as indicated. Scale bars: 100 μm. Student’s t test and ANOVA compared the difference in A, B, D and F. Log-rank test for survival comparison in F. * P < 0.05 vs. LV-sh-Scb. Data are shown as mean ± s.e.m. (error bars) in A, B, D and F.

Figure 8.

Identification of emetine as an inhibitor blocking E2F4-USP2 interaction. (A) Venn diagram indicating identification of 327 potential compounds interacting with USP2 derived from DINES (https://www.genome.jp/tools/dinies/) database, and overlapping analysis with chemicals inducing upregulation or downregulation of target genes (ATG2A and ULK2) using ChIP-X program. (B) Co-IP and Western blot assays showing the interaction of USP2 with E2F4 in gastric cancer AGS cells treated with seven potential compounds as indicated. (C) Representative images (left panel) and quantification (right panel) of BiFC assay revealing the direct interaction between E2F4 with USP2 in AGS cells treated with chemicals as indicated. Scale bars: 10 μm. (D) Dual-luciferase assay indicating the E2F4 transactivation activity in AGS cells treated with chemicals as indicated. (E) Co-IP and Western blot assays showing the interaction of USP2 with E2F4, ATG2A, ULK2, CCND1, or MDM2 in gastric cancer AGS cells treated PBS or emetine (EMT, 0.5 µmol/L) for 48 h. ANOVA compared the difference in C and D. * P < 0.05 vs. PBS. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in B-E.

Figure 9.

Emetine suppresses autophagy and cancer progression via blocking E2F4-USP2 interaction. (A) Denatured IP and Western blot assays indicating the ubiquitination of E2F4 in AGS cells treated with emetine (EMT, 0.5 µmol·L⁻¹), and those transfected with empty vector (mock) or USP2. (B) Western blot assay revealing the E2F4 levels in MGC-803 and AGS cells treated with different doses of EMT, and those co-treated with MG132 (10 μmol·L⁻¹). (C) Western blot assay showing the levels of LC3-I, LC3-II and BECN1/Beclin 1 in MGC-803 and AGS cells treated with different doses of EMT. (D) Representative images (upper panel) and quantification (lower panel) of RFP-GFP-LC3 and FluoZin-3 reporters in AGS cells treated with EMT (0.5 µmol·L⁻¹) for 48 h. Scale bars: 10 μm. (E) MTT colorimetric assay depicting changes in viability of AGS cells treated with different doses of EMT for 24 h or EMT (0.5 µmol·L⁻¹) for time points as indicated. (F and G) Representative images (left panel) and quantification (right panel) of soft agar (F) and matrigel invasion (G) assays indicating Anchorage-independent growth and invasion of MGC-803 and AGS cells stably transfected with mock or E2F4, and those treated with EMT (0.5 µmol·L⁻¹). (H) Representative images (left upper panel), tumor growth curve (left lower panel), weight (right upper panel), relative zinc ion levels (right upper panel), and immunohistochemical staining of MKI67/Ki-67 and PECAM1/CD31 (right lower panel) of subcutaneous xenografts formed by AGS cells in nude mice that received intraperitoneal administration of EMT (10 mg/kg, n = 5 for each group), with body weight (left lower panel) as indicated. Scale bars: 100 μm. (I) Representative images (upper panel), HE staining (middle panel), quantification of lung metastatic colonization (lower left panel), and survival curves (lower right panel) of nude mice treated with tail vein injection of AGS cells and PBS or EMT (10 mg/kg, n = 5 for each group). Scale bars: 100 μm. Student’s t test and ANOVA compared the difference in D-I. Log-rank test for survival comparison in I. * P < 0.05 vs. PBS or PBS+mock. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in A-G.

Figure 10.

USP2, E2F4, and target genes are associated with outcome of gastric cancer. (A) Western blot assay indicating the levels of USP2, E2F4, LC3-I, and LC3-II in gastric cancer tissues (T) or their normal counterparts (N). (B) Real-time qRT-PCR assay showing the levels (normalized to ACTB/β-actin) of USP2, E2F4, ATG2A, and ULK2 in gastric cancer tissues and their normal counterparts (n = 21). (C) Kaplan-Meier curves indicating the overall (OS) and first progression (FP) survival of gastric cancer cases derived from KM Plotter (http://kmplot.com) with high or low expression of USP2 (cutoff values = 3.46 and 3.00), ATG2A (cutoff values = 8.59 and 8.42), or ULK2 (cutoff values = 5.73 and 5.55). (D) Pearson’s coefficient correlation analysis depicting the positive expression correlation of E2F4 with ATG2A or ULK2 in gastric cancer tissues (n = 875). (E) The mechanisms underlying the roles of USP2 and E2F4 in autophagy and cancer progression: as a deubiquitinating enzyme, USP2 interacts with and stabilizes E2F4 protein, resulting in transcriptional upregulation of autophagic genes ATG2A and ULK2, autophagic degradation of metallothioneins, decrease of labile intracellular zinc ions, and increase of tumorigenesis and aggressiveness. Meanwhile, emetine is able to block USP2-E2F4 interaction and suppress cancer progression. Student’s t test compared the difference in B. Log-rank test for survival comparison in C. Pearson’s correlation analysis in D. * P < 0.05 vs. normal. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in A and B.