Leptin-induced epithelial-mesenchymal transition in breast cancer cells requires β-catenin activation via Akt/GSK3- and MTA1/Wnt1 protein-dependent pathways

Abstract

Perturbations in the adipocytokine profile, especially higher levels of leptin, are a major cause of breast tumor progression and metastasis; the underlying mechanisms, however, are not well understood. In particular, it remains elusive whether leptin is involved in epithelial-mesenchymal transition (EMT). Here, we provide molecular evidence that leptin induces breast cancer cells to undergo a transition from epithelial to spindle-like mesenchymal morphology. Investigating the downstream mediator(s) that may direct leptin-induced EMT, we found functional interactions between leptin, metastasis-associated protein 1 (MTA1), and Wnt1 signaling components. Leptin increases accumulation and nuclear translocation of β-catenin leading to increased promoter recruitment. Silencing of β-catenin or treatment with the small molecule inhibitor, ICG-001, inhibits leptin-induced EMT, invasion, and tumorsphere formation. Mechanistically, leptin stimulates phosphorylation of glycogen synthase kinase 3β (GSK3β) via Akt activation resulting in a substantial decrease in the formation of the GSK3β-LKB1-Axin complex that leads to increased accumulation of β-catenin. Leptin treatment also increases Wnt1 expression that contributes to GSK3β phosphorylation. Inhibition of Wnt1 abrogates leptin-stimulated GSK3β phosphorylation. We also discovered that leptin increases the expression of an important modifier of Wnt1 signaling, MTA1, which is integral to leptin-mediated regulation of the Wnt/β-catenin pathway as silencing of MTA1 inhibits leptin-induced Wnt1 expression, GSK3β phosphorylation, and β-catenin activation. Furthermore, analysis of leptin-treated breast tumors shows increased expression of Wnt1, pGSK3β, and vimentin along with higher nuclear accumulation of β-catenin and reduced E-cadherin expression providing in vivo evidence for a previously unrecognized cross-talk between leptin and MTA1/Wnt signaling in epithelial-mesenchymal transition of breast cancer cells.

FIGURE 1.

Leptin induces epithelial-mesenchymal transition, tumorsphere formation, migration, and invasion of breast cancer cells. A, MCF7 cells were serum-starved for 16 h followed by treatment with 100 ng/ml leptin (L) or were untreated (U) for 3 days. Morphological changes associated with EMT are shown in phase-contrast images. The presence of spindle-shaped cells, increased intercellular separation, and pseudopodia were noted in treated cells but not in untreated cells. B, MCF7 cells were treated as in A. Fluorescent confocal microscopy showed the alterations in the cytoskeleton by leptin treatment. Reorganization of actin into pseudopodia and stress fibers were noted in leptin-treated cells. Nuclei were visualized with DAPI staining. C, MCF7 cells were treated with 100 ng/ml leptin (L) or were untreated (U). Total lysates were immunoblotted for fibronectin, N-cadherin, vimentin, Snail, Slug, Occludin, and E-cadherin expression levels. Leptin treatment increased fibronectin, N-cadherin, vimentin, Snail, and Slug expression, whereas E-cadherin and occludin expression was decreased. Actin was used as control. D, densitometric analysis was performed on multiple blots for each target protein. The histogram is the mean of densitometric analysis showing relative density units (RDU) of the Western blot signal for target proteins normalized to actin. *, p < 0.005 compared with untreated controls. E, MCF7 cells were subjected to the scratch-migration assay in the absence (U) or presence of 100 ng/ml leptin (L) for 24 h. Fold-change in migration is shown in bar graphs. *, p < 0.005, compared with untreated controls. F, MCF7 cells were cultured in Matrigel invasion chambers followed by treatment with leptin for 24 h. The number of cells that invaded through the Matrigel was counted in five different regions. The slides were blinded to remove counting bias. The results show mean of three independent experiments performed in triplicate. *, p < 0.005 compared with untreated controls (U). G, MCF7 cells were subjected to tumorsphere formation in the absence (U) or presence of 100 ng/ml leptin (L) as indicated. Number of tumorsphere was counted in five different views under a microscope. The results are presented as fold-change with untreated. *, p < 0.005, compared with untreated controls.

FIGURE 2.

Leptin increases expression, nuclear accumulation, and recruitment of β-catenin to responsive gene promoters. A, MCF7; B, MDA-MB-231 cells were treated with 100 ng/ml leptin for indicated time intervals. Total lysates were immunoblotted for β-catenin expression. The membranes were re-blotted using actin antibody as control. The blots are representative of multiple independent experiments. The histogram is the mean of densitometric analysis showing relative density units (RDU) of the Western blot signal for β-catenin normalized to actin. *, p < 0.005 compared with untreated controls. Leptin (L) increases the expression of β-catenin. C and D, breast cancer cells were treated with 100 ng/ml leptin for the indicated time intervals followed by nuclear (Nucl) and cytoplasmic (Cyto) extract isolation. Equal amounts of proteins were immunoblotted using antibodies against β-catenin. The membranes were re-blotted using tubulin, histone, and actin antibodies as controls. The blots are representative of multiple independent experiments. The histogram is the mean of densitometric analysis showing relative density units (RDU) of the Western blot signal for β-catenin normalized to either tubulin or histone. *, p < 0.005 compared with untreated controls. Leptin treatment induces nuclear accumulation of β-catenin. E, soluble chromatin was prepared from breast cancer cells treated and untreated with leptin as indicated and immunoprecipitated with 5 μg of specific antibodies against β-catenin overnight at 4 °C. The immune complexes were pulled down with protein A-agarose/salmon sperm DNA beads and washed extensively as described under “Experimental Procedures,” and cross-linking was reversed. The purified DNA was analyzed by PCR using primers spanning the TCF/LEF site at cyclin D1 and fibronectin promoter. EBAG9 was used as a control gene. Leptin increases recruitment of β-catenin to responsive gene promoters.

FIGURE 3.

Leptin elevates phosphorylation of GSK3β and decreases the formation of GSK3β-LKB1-Axin complex. A, MCF7 breast cancer cells were treated with 100 ng/ml leptin for indicated time intervals. Total lysates were immunoblotted using specific antibodies for phospho-GSK3β. The membranes were re-blotted using total GSK3β and actin antibodies as control. The blots are representative of multiple independent experiments. The histogram is the mean of densitometric analysis showing relative density units (RDU) of the Western blot signal for phosphorylated GSK3β normalized to total GSK3β. *, p < 0.005 compared with untreated controls. Leptin treatment increases phosphorylation of GSK3β. B, breast cancer cells were untreated (U) or treated with 100 ng/ml leptin (L) and subjected to immunoprecipitation (IP) analysis using specific antibodies against LKB1 and Axin. The immunoprecipitates were analyzed by immunoblot analysis using anti-GSK3β antibodies. Immunoprecipitation with IgG was used as control. Leptin decreases the formation of the GSK3β-LKB1-Axin complex. C, breast cancer cells were treated with 100 ng/ml leptin alone (L) and in combination with LiCl (L+Li). Untreated cells are denoted with U. Total lysates were immunoblotted using antibodies against phospho-GSK3β and β-catenin. The membranes were re-blotted using total GSK3β and actin antibodies as control. The blots are representative of multiple independent experiments. Combined treatment with leptin and LiCl further increases β-catenin expression. WB, Western blot.

FIGURE 4.

Leptin-mediated increase in GSK3β phosphorylation involves activation of Akt as well as Wnt1. A, breast cancer cells were treated with 100 ng/ml leptin for indicated time intervals. Total lysates were immunoblotted for phospho-Akt expression. The membranes were re-blotted using total Akt and actin antibodies as control. The blots are representative of multiple independent experiments. The histogram is the mean of densitometric analysis showing relative density units (RDU) of the Western blot signal for phosphorylated Akt normalized to total Akt. *, p < 0.001 compared with untreated controls. Leptin treatment increases phosphorylation of Akt. B, breast cancer cells were untreated (U) or treated with 100 ng/ml leptin alone (L), 10 μm Akt inhibitor LY294002 alone (LY), and in combination (L+LY). Total lysates were immunoblotted for phospho-GSK3β and phospho-Akt expression levels. The membranes were re-blotted using total GSK3β, total Akt, and actin antibodies as control. LY294002 reduces leptin-induced phosphorylation of GSK3β. C, breast cancer cells were transiently transfected with siAkt#1 or siAkt#2 siRNAs (Invitrogen) using FuGENE (Invitrogen). Immunoblot analysis was performed using Akt antibodies, and actin antibodies were used as control. D, breast cancer cells were transfected with siAkt#2 for 24 h followed by leptin treatment for 6 h. Total lysates were subjected to immunoblot analysis using specific antibodies for phospho-GSK3β, fibronectin, and vimentin. The membranes were re-blotted using actin. The blots are representative of multiple independent experiments. Akt silencing inhibits leptin-induced GSK3β phosphorylation, fibronectin, and vimentin expression. E, breast cancer cells were treated with 100 ng/ml leptin for indicated time intervals. Total lysates were immunoblotted using anti-Wnt1 antibodies. Immunoblots for actin levels were used as control. The blots are representative of multiple independent experiments. The histogram is the mean of densitometric analysis showing relative density units (RDU) of the Western blot signal for Wnt1 normalized to actin. *, p < 0.005 compared with untreated controls. Leptin increases Wnt1 expression. F, breast cancer cells were transiently transfected with siWnt1#1 or siWnt1#2 siRNAs (Invitrogen) using FuGENE (Invitrogen). Immunoblot analysis was performed using Wnt1 antibodies, and actin antibodies were used as control. G, breast cancer cells were transfected with siWnt1#1 for 24 h followed by leptin treatment for 6 h. Total lysates were subjected to immunoblot analysis using specific antibodies for phospho-GSK3β. The membranes were re-blotted using total GSK3β. The blots are representative of multiple independent experiments. Wnt1 inhibition inhibits leptin-induced GSK3β phosphorylation.

FIGURE 5.

Leptin induces the expression of MTA1 that plays an integral role in leptin-induced GSK3β phosphorylation, β-catenin accumulation, and nuclear translocation. A, MCF7 breast cancer cells were treated with 100 ng/ml leptin for indicated time intervals. Total protein lysates were immunoblotted using specific antibodies for MTA1. Immunoblot for actin levels was used as control. The blots are representative of multiple independent experiments. The histogram is the mean of densitometric analysis showing relative density units (RDU) of the Western blot signal for MTA1 normalized to actin. *, p < 0.001 compared with untreated controls. Leptin increases MTA1 expression in breast cancer cells. B, MCF7 cells were transiently transfected with siMTA1#1 or siMTA1#2 siRNAs (Invitrogen) using FuGENE (Invitrogen). Immunoblot analysis was performed using MTA1 antibodies, and actin antibodies were used as control. C, breast cancer cells were transfected with siMTA1#1 for 24 h followed by leptin treatment for 6 h. Total lysates were immunoblotted using antibodies against Wnt1, phospho-GSK3β, β-catenin, fibronectin, and vimentin along with total GSK3β and actin. The blots are representative of multiple independent experiments. Inhibition of MTA1 inhibits leptin-induced GSK3β phosphorylation, β-catenin, fibronectin, and vimentin expression. D, breast cancer cells were transfected with siMTA1#1 for 24 h followed by leptin treatment for 6 h (siMTA1+L), untreated (U) or treated with leptin alone (L) for 6 h and subjected to immunofluorescence analysis of β-catenin. MTA1 silencing inhibits leptin-induced nuclear localization of β-catenin. E, densitometric analysis was performed on multiple blots for each target protein. The histogram is the mean of densitometric analysis showing relative density units (RDU) of the Western blot signal for target proteins (Wnt1, p-GSK3β, β-catenin, fibronectin, and vimentin) normalized to actin or unphosphorylated total protein. *, p < 0.001 compared with untreated controls. Inhibition of MTA1 inhibits leptin-induced GSK3β phosphorylation, β-catenin, fibronectin, and vimentin expression.

FIGURE 6.

Depletion of β-catenin or inhibition of β-catenin activity using ICG-001 abrogates leptin-mediated tumorsphere formation, invasion, and epithelial-mesenchymal transition of breast cancer cells. A, β-catenin was depleted in MCF7 cells using lentiviral β-catenin shRNA constructs and a negative control construct that was created in the same vector system (pLKO.1). Stable pools of β-catenin-depleted (p-β-cat^sh) and vector control (pLKO.1) cells were used for total protein isolation, and equal amounts of proteins were subjected to immunoblot analysis using β-catenin antibodies. The membranes were re-blotted using actin antibody as control. The blots are representative of multiple independent experiments. B, parental, pLKO.1, and p-β-cat^sh cells were subjected to tumorsphere formation in the presence (L) or absence (U) of 100 ng/ml leptin. The number of tumorspheres was counted in 10 different views under microscope. The results are presented as fold-change with untreated. *, p < 0.001, compared with untreated controls. Silencing of β-catenin abrogates leptin-induced tumorsphere formation. C, parental, pLKO.1, and p-β-cat^sh cells were subjected to Matrigel-invasion assay in the presence (L) or absence (U) of 100 ng/ml leptin. The number of cells that invaded through the Matrigel was counted in five different regions. The slides were blinded to remove counting bias. The results show mean of three independent experiments performed in triplicate. *, p < 0.005 compared with untreated controls (U). D, parental, pLKO.1, or p-β-cat^sh cells were treated with 100 ng/ml leptin (L) or untreated (U). Equal amounts of proteins were subjected to immunoblot analysis using fibronectin, vimentin, and E-cadherin antibodies. The membranes were re-blotted using actin antibody as control. The blots are representative of multiple independent experiments. Depletion of β-catenin abrogates leptin-mediated increase in mesenchymal markers and decrease in epithelial markers. E, MCF7 cells were treated with 5 μm ICG-001 followed by total RNA isolation. ICG-001 treatment reduced cyclin D1 expression. Actin was used as loading control. F, MCF7 cells were untreated (U), treated with 100 ng/ml leptin (L), and 5 μm ICG-001 (ICG001) alone or in combination and observed for morphological changes associated with EMT as shown in phase-contrast images. Leptin-treated cells exhibited spindle-shaped cells, increased intercellular separation, and pseudopodia. These alterations were absent in ICG-001 + leptin-treated cells, which closely resembled the untreated cells. G, MCF7 cells were treated as in E followed by total RNA isolation and RT-PCR using primers specific for Snail, Zeb2, Slug, and Actin. ICG-001 treatment inhibited leptin-induced expression of Snail, Zeb2, and Slug. Actin expression level was used as control. H, MCF7 cells were transiently transfected with si-β-catenin or si-control siRNAs (Invitrogen) using FuGENE (Invitrogen). Immunoblot analysis was performed using β-catenin antibodies, and actin antibodies were used as control. si-β-Catenin inhibits the expression of β-catenin. Breast cancer cells transfected with si-β-catenin were treated with leptin and followed for morphological changes. Leptin treatment did not induce any morphological changes in cells lacking β-catenin expression.

FIGURE 7.

In vivo evidence for leptin-mediated alterations in EMT markers as well as Wnt1, GSK3β, β-catenin, vimentin, and Ki-67 expression levels. A, MDA-MB-231 cell-derived tumors were developed in nude mice and treated with vehicle (V) or leptin (L) for 6 weeks. Tumors were collected at the end of 6 weeks and subjected to immunohistochemical analysis using GSK3β, Wnt1, β-catenin, vimentin, and Ki-67 antibodies. Bar diagrams show quantitation of protein expression in tumors from vehicle and leptin-treated mice. Columns, mean (n = 8); bar, S.D. * indicates significantly different (p < 0.005) compared with control. B, schematic model of leptin-stimulated Akt/GSK3β and MTA1/Wnt1 axes activating β-catenin leading to epithelial-mesenchymal transition in breast cancer cells. Under basal conditions, in the absence of leptin, unphosphorylated GSK3-β forms a destruction complex with LKB1, Axin, and APC leading to phosphorylation, ubiquitination, and degradation of β-catenin resulting in repression of target genes. Leptin stimulation leads to Akt phosphorylation, which in turn phosphorylates GSK3-β rendering it inactive, which leads to disintegration of destruction complex. Leptin also stimulates MTA1 overexpression leading to increased Wnt1 expression that phosphorylates GSK3-β. These events lead to accumulation, nuclear translocation of β-catenin, and activation of target genes. Leptin stimulates β-catenin activation via Akt/GSK3-dependent and MTA1/Wnt1-dependent pathways.