PTEN phosphatase inhibits metastasis by negatively regulating the Entpd5/IGF1R pathway through ATF6

Abstract

PTEN encodes a tumor suppressor with lipid and protein phosphatase activities whose dysfunction has been implicated in melanomagenesis; less is known about how its phosphatases regulate melanoma metastasis. We demonstrate that PTEN expression negatively correlates with metastatic progression in human melanoma samples and a PTEN-deficient mouse melanoma model. Wildtype PTEN expression inhibited melanoma cell invasiveness and metastasis in a dose-dependent manner, behaviors that specifically required PTEN protein phosphatase activity. PTEN phosphatase activity regulated metastasis through Entpd5. Entpd5 knockdown reduced metastasis and IGF1R levels while promoting ER stress. In contrast, Entpd5 overexpression promoted metastasis and enhanced IGF1R levels while reducing ER stress. Moreover, Entpd5 expression was regulated by the ER stress sensor ATF6. Altogether, our data indicate that PTEN phosphatase activity inhibits metastasis by negatively regulating the Entpd5/IGF1R pathway through ATF6, thereby identifying novel candidate therapeutic targets for the treatment of PTEN mutant melanoma.

Keywords: Cancer; Cell biology; Functional aspects of cell biology.

Graphical abstract

Figure 1

PTEN expression is correlated to metastatic behavior in melanoma (A) PTEN loss led to markedly reduced survival in the UV-irradiated HGF transgenic mouse model. HGF^Tg/+/PTEN^fl/fl/Cre^Tg/+ and HGF^Tg/+/PTEN^fl/fl/Cre^+/+ mice were developed by crossing the HGF^Tg/+, PTEN^fl/fl and Tyr:CreER^Tg/Tg GEM models. Melanomas were initiated by a single dose of UV radiation at 3.5 days of age. The depletion of PTEN was achieved by administration of 4-hydroxytamoxifen (4-OHT) at 8 weeks of age. Kaplan-Meier survival analysis of 4-OHT-treated HGF^Tg/+/PTEN^fl/fl/Cre^Tg/+ (n = 13), HGF^Tg/+/PTEN^fl/fl/Cre^+/+ (n = 12), and corn oil treated-HGF^Tg/+/PTEN^fl/fl/Cre^Tg/+ (n = 8) mice. Log-rank tests of survival plots of the data indicated a statistically significant difference between the following survival curves: HGF^Tg/+/PTEN^fl/fl/Cre^Tg/+ with 4-OHT vs. HGF^Tg/+/PTEN^fl/fl/Cre^Tg/+ with corn oil (p<0.0001) and HGF^Tg/+/PTEN^fl/fl/Cre^Tg/+ with 4-OHT vs. HGF^Tg/+/PTEN^fl/fl/Cre^+/+ with 4-OHT (p<0.0001). There was no statistically significant difference between HGF^Tg/+/PTEN^fl/fl/Cre^Tg/+ with corn oil vs. HGF^Tg/+/PTEN^fl/fl/Cre^+/+ with 4-OHT. (B) Incidence of metastasis in mice. PTEN-deficient UV-irradiated HGF^tg/+ mice suffered a significantly higher risk of metastasis than PTEN-WT UV-irradiated HGF^tg/+ mice, Fisher exact test p = 0.0248. (C) Representative lung image with metastatic lesion from HGF^Tg/+/PTEN^fl/fl/Cre^Tg/+ mouse with corn oil. (D) PTEN expression and gross pulmonary metastases from a panel of well-established human melanoma A375p (p), A375sm (sm), A375c5 (c5) and A375c28 (c28) cell lines. (a) PTEN protein level by western blot; the quantitated data of the fold changes normalized with β-actin; (b) gross pulmonary metastases by tail-vein injection. In comparison with p cells, sm and c28 cells had significantly increased pulmonary metastatic potential (p <0.001). (E) Knockdown of endogenous PTEN by shRNA enhanced metastatic potential in A375p cells. (a) three stable cell clones expressing reduced PTEN to different degrees by 43%, 22% and 69%, respectively; (b) gross pulmonary metastases. Compared to control c, three clones showed variously increased pulmonary metastasis; clone #6 and #18 had a statistically significant increase in metastasis (p<0.001). (F) Overexpression of PTEN in human melanoma A375sm and A375c28 cells significantly inhibited pulmonary metastases (p < 0.001). C, empty vector; PTEN, overexpression of PTEN. (G) Overexpression of PTEN in mouse melanoma B16F1 cells significantly inhibited pulmonary metastases (p = 0.0033) (a), whereas knockdown of endogenous PTEN enhanced pulmonary metastases significantly (p = 0.00019) (b). C, empty vector; WT, overexpression of PTEN; shRNA, knockdown of endogenous PTEN by shRNA for PTEN. (H) Representative H&E stained lung sections are shown for the experimental metastases from a panel of well-established human melanoma A375p (p), A375sm (sm), A375c5 (c5) and A375c28 (c28) cell lines (D) (scale bar = 5 mM). Data represented as mean ± SEM for all columns. The p value is shown by an unpaired t-test (two-tailed).

Figure 2

PTEN regulates metastasis through its phosphatase activity independent of lipid phosphatase activity (A and B) PTEN phosphatase dead mutant enhanced metastasis. Western blot analysis of lysates from B16F1 (A, top panel) or 37-7 cell (B, top panel) transfectants harboring empty vector (c), PTEN wildtype (WT or GFP-PTEN) and PTEN phosphatase dead mutant (ΔLP or GFP-PTEN ΔLP) and gross pulmonary metastases in different hosts (FVB host with 1x10e6 cells, nude host with 1x10e5 cells) in vivo (bottom panel of A or B). c, empty vector control; 1, 2 and 3 in 37-7 cells as three different clones. (C) PTEN lipid phosphatase mutant was not required for suppression of metastasis. Enforced expression of various PTEN mutants in B16F1 cells was represented by western blot (top panel) and gross pulmonary metastases from B16F1 cell transfectants (bottom panel). C, empty vector; WT, PTEN wildtype; ΔL, PTEN lipid phosphatase deficient; ΔLP, PTEN phosphatase dead mutant. (D) Representative histopathology (H&E staining) of lung sections with metastases from mice bearing B16F1 cells with empty vector (c), PTEN WT (WT), PTEN lipid phosphatase deficient (ΔL), PTEN phosphatases dead mutant (ΔLP) (scale bar = 5 mM). (E–G) Phenotypic effects of various PTEN mutants in B16F1 in vitro. The proliferation rates (E) of cells were assessed by measuring the incorporation of [³H] thymidine. Motility (F) and invasiveness (G) were determined using Transwell culture plates for 12 or 48 h, respectively. C, vector alone; WT, PTEN wildtype; ΔL, PTEN lipid phosphatase deficient; ΔLP, PTEN phosphatase dead mutant. B16F1 cells transfected with PTEN wildtype (WT) demonstrated significant inhibition in growth (p = 0.0012), motility (p = 0.0073) and invasiveness (p = 0.013) in vitro compared with empty vector control (c). The cells expressing PTEN phosphatase dead mutant (ΔLP) or PTEN lipid phosphatase deficient (ΔL) showed a significant increase in cell growth (p<0.05), but no significant difference in cell motility. However, the cells with PTEN phosphatase dead mutant (ΔLP) showed an increase in cell invasiveness (p = 0.008). Data is representative of three independent experiments. Graphs show the mean ± SEM. The p value is shown by an unpaired t-test (two-tailed).

Figure 3

PTEN protein phosphatase inhibits early survival of tumor cells that reached the lung and mediates early metastatic survival in vivo (A) Images of fluorescence labeled single tumor cells arriving in the lung after experimental metastasis by the single cell-metastasis imaging system. EV, empty vector; WT, wildtype PTEN; ΔL, lipid phosphatase mutant; ΔLP, PTEN phosphatase dead mutant. (B) Quantitative data of fluorescence labeled single tumor cell arriving in lung after experimental metastasis (each group counted 5 samples). PTEN phosphatase dead mutant (ΔLP) significantly enhanced early survival of tumor cells that reached the lung after injection 24 h (ΔLP vs WT, p = 0.01; ΔLP vs ΔL, p = 0.011, no significant difference between ΔL vs WT). Data represented as mean ± SEM for all columns. The p value is shown by an unpaired t-test (two-tailed).

Figure 4

The gene expression patterns associated with melanoma progression and identification of Entpd5 as a downstream target of PTEN phosphatase (A) The significantly expressed genes in B16F1 cells with PTEN wildtype (WT), lipid phosphatase mutant (ΔL) or phosphatase dead mutant (ΔLP) forms compared with empty vector control (c) were filtered by a p value of 0.05 and absolute value of fold change of 1.5 in ANOVA analysis. (B) The common significantly altered gene list between PTEN wildtype (WT), lipid phosphatase mutant (ΔL) and phosphatase dead mutant (ΔLP) in B16F1 cells based on hierarchical clustering. (C and D) Validation of genes identified in cDNA microarray analysis by Quantitative RT-PCR (qRTPCR) (C) and Western blot analysis (D). β-actin was used as a control.

Figure 5

Entpd5 regulates tumor metastasis in melanoma (A) Knockdown of Entpd5 blocks PTEN phosphatase dead mutant-stimulated metastasis in melanomas. (A-a) Western blot analysis of stable B16F1-PTEN ΔLP cells transfected with shRNA plasmids for Entpd5. (A-b,c,d) Knockdown of Entpd5 affects cell growth (A-b), cell motility (A-c) and cell invasiveness (A-d) in vitro. Data is representative of three independent experiments. Graphs show the mean ± SEM. The p value is shown by an unpaired t-test (two-tailed). (A-e) Gross pulmonary metastases from stable B16F1-PTEN ΔLP cells transfected with shRNA plasmids for Entpd5. c, empty vector; sh1,sh2, sh3 and sh4 represent four different shRNA forms of mouse Entpd5. Data represented as mean ± SEM for all columns. The p value is shown by an unpaired t-test (two-tailed). (A-f) Representative histopathology (H&E staining) of lung sections with metastases from mice bearing B16F1-PTEN ΔLP cells transfected with shRNA plasmids for Entpd5 (scale bar = 5 mM). c, empty vector; sh1,sh2, sh3 and sh4, four different shRNA forms of Entpd5. (B) Blocking endogenous Entpd5 by antisense results in significant inhibition of metastasis. (B-a) Western blot analysis of stable A375sm cells transfected with antisense of Entpd5-expressing plasmid. (B-b, c, d) Alteration of cell growth (B-b), motility (B-c) as well as invasiveness (B-d) in vitro in human A375sm cells transfected with antisense of Entpd5 expressing plasmid. Data is representative of three independent experiments. Graphs show the mean ± SEM. The p value is shown by an unpaired t-test (two-tailed). (B-e) Gross pulmonary metastases from A375sm cell transfected with Entpd5 antisense. c, empty vector; As, antisense for Entpd5. Data represented as mean ± SEM for all columns. The p value is shown by an unpaired t-test (two-tailed). (C) Overexpression of Entpd5 abrogates PTEN wildtype-mediated metastatic suppression. (C-a) Western blot analysis of stable B16F1-PTEN WT and A375p cell lines transfected with an Entpd5-expressing plasmid. (C-b, c, d) Forced expression of Entpd5 stimulates cellular proliferation (C-b), motility (C-c) and invasiveness (C-d) in vitro. Data is representative of three independent experiments. Graphs show the mean ± SEM. The p value is shown by an unpaired t-test (two-tailed). (C-e, f) Gross pulmonary or liver metastases from B16F1-PTEN and A375p cells transfected with an Entpd5 expressing vector. – or c, empty vector; + or Entpd5; Entpd5 expressing vector. Data represented as mean ± SEM for all columns. The p value is shown by an unpaired t-test (two-tailed).

Figure 6

IGF1R regulates tumor metastasis in melanoma cells (A) Knockdown of IGF1R expression also blocks PTEN phosphatase dead mutant-stimulated metastasis. (A-a) Western blot analysis of stable B16F1-PTEN ΔLP cells transfected with shRNA plasmids for mouse IGF1R. (A-b) Gross pulmonary metastases from B16F1-PTEN ΔLP cells transfected with shRNA for mouse IGF1R. Data represented as mean ± SEM for all columns. The p value is shown by an unpaired t-test (two-tailed). (A-c) Effects on cell growth by shRNA of IGF1R. Data is representative of three independent experiments. Graphs show the mean ± SEM. The p value is shown by an unpaired t-test (two-tailed). c, empty vector; sh1 and sh2, shRNA expression plasmid for mouse IGF1R. (B) Knockdown of IGF1R expression inhibits metastasis in human melanoma A375sm cells. (B-a) Western blot analysis of stable A375sm cells transfected with shRNA plasmids for human IGF1R. (B-b) Gross pulmonary metastases from transfected cells. Data represented as mean ± SEM for all columns. The p value is shown by an unpaired t-test (two-tailed). (B-c) Effects on cell growth by shRNAs of IGF1R. Data is representative of three independent experiments. Graphs show the mean ± SEM. The p value is shown by an unpaired t-test (two-tailed). c, empty vector; sh1 and sh2, shRNA expressing plasmid for human IGF1R. (C) Forced expression of IGF1R results in the abrogation of PTEN-mediated inhibition of metastasis. (C-a) Western blot analysis of stable B16F1-PTEN cells transfected with IGF1R expressing plasmid. (C-b) Gross pulmonary metastases from transfected cells. Data represented as mean ± SEM for all columns. The p value is shown by an unpaired t-test (two-tailed). (C-c) Effects on cell growth by IGF1R. Data is representative of three independent experiments. Graphs show the mean ± SEM. The p value is shown by an unpaired t-test (two-tailed). c, empty vector; IGF1R, IGF1R expressing plasmid.

Figure 7

PTEN expression negatively correlates with Entpd5 and IGF1R in human melanoma samples (A) Immunohistochemistry analyses of tissue microarrays were performed using antibodies against PTEN, Entpd5 and IGF1R in consecutively cut sections of nevi (n = 18), melanomas (n = 56) (stage IB, n = 6; Stage II, n = 26; Stage III, n = 6, other, n = 18) and metastatic melanomas (n = 26). Representative examples are shown of tumor cores with high-, medium-, low- or none-intensity PTEN staining, with their corresponding intensities of Entpd5 and IGF1R (low-, medium- or high-intensity staining). Scale bar = 20 μM. (B) Representative example showing a significantly negative correlation between PTEN and Entpd5 at the transcription level in the GSE7929 dataset by Pearson r (two-tailed) analysis. (C-E and I) PTEN (C), Entpd5 (D), IGF1R (E) and ATF6 (I) expression was determined by immunohistochemistry using antibodies against PTEN, Entpd5, IGF1R or ATF6 in tissue microarray with 30 melanomas and 30 metastatic melanomas. They were scored for a combination of staining intensity on a 0–3 scale (none =0, low = 1, medium = 2 and high = 3) and a frequency of tumor labeling on a 0–5 scale (no cells = 0, <10% = 1, 10–32% = 2, 33–65%= 3, 66–99% = 4 and 100% = 5). (F, G) Expression of PTEN and Entpd5 in poorly and highly metastatic melanoma tumors in the GSE7929 dataset. Data represented as mean ± SEM. The p value is shown by an unpaired t-test (two-tailed). (H) Analyses of GSE53118 and GSE54467 late-stage melanoma datasets show that high PTEN with low Entpd5 expression is associated with a longer time of survival in late-stage malignant melanoma [p = 0.0293 by Logrank (Mantel-Cox) test].

Figure 8

ATF6 regulated by PTEN can directly bind to the Entpd5 promoter and regulate the expression of Entpd5 (A) ATF6 protein expression was analyzed by western blot in cells bearing various PTEN mutants. PTEN WT expression decreased ATF6 expression, whereas PTEN ΔLP increased ATF6 expression. (B) Effects of PTEN mutants on 5xATF6 binding site-driven luciferase activity. Data is representative of three independent experiments. Graphs show the mean ± SEM. (C and D) Entpd5 and IGF1R protein expression was analyzed by western blot in B16F1 cell lines transfected with either an ATF6 expression vector or treated with siRNA for ATF6. Knockdown of ATF6 through an RNAi mechanism inhibited Entpd5 expression in B16F1-PTEN ΔLP (C), whereas ectopic ATF6 expression stimulated Entpd5 expression in B16F1-PTEN (D). c, empty vector control. (E) Three candidate ATF6 binding sites consisting of the 5′-CCAC[GA]-3′ half of the ER stress response element (ERSE) were found within the 5′-flanking region (between −2342 and -1bp) of the Entpd5 promoter. (F) The physical interaction of ATF6 to Entpd5 gene promoter was assessed by ChIP assay. Native ATF6 in B16F1-PTEN ΔLP cells was analyzed using an anti-ATF6 antibody. ATF6 bound to the −840 to −641 and −222 to −25 regions of the Entpd5 promoter, but not the −2827 to −2627 regions. Inp, input; M, markers. (G), Entpd5 gene promoter (−1010 to −1) activity was shown to be responsive to increasing amounts of an ATF6 expression vector using a firefly luciferase (Luc) reporter, while not to be responsive to an ATF6 mutant form that lost its binding activity. (H) Using the same luciferase assay, the addition of siRNA for ATF6 with an ATF6-expression vector inhibited luciferase activity driven by the Entpd5 promoter. (I) Entpd5 gene promoter (−1010 to −1) activity was shown to be stimulated by increasing amounts of a PTEN ΔLP expression vector using a firefly luciferase (Luc) reporter, while not to be responsive to PTEN ΔL. (J) Using the same luciferase assay, addition of siRNA for ATF6 inhibited PTEN ΔLP-stimulated luciferase activity driven by the Entpd5 promoter in a co-transfection assay. Data is representative of three independent experiments. Graphs show the mean ± SEM.