ERα/LKB1 complex upregulates E‐cadherin expression and stimulates breast cancer growth and progression upon adiponectin exposure

Adiponectin is the major adipocytes‐secreted protein involved in obesity‐related breast cancer growth and progression. We proved that adiponectin promotes proliferation in ERα‐positive breast cancer cells, through ERα transactivation and the recruitment of LKB1 as ERα‐coactivator. Here, we showed that adiponectin‐mediated ERα transactivation enhances E‐cadherin expression. Thus, we investigated the molecular mechanism through which ERα/LKB1 complex may modulate the expression of E‐cadherin, influencing tumor growth, progression and distant metastasis. We demonstrated that adiponectin increases E‐cadherin expression in ERα‐positive 2D and higher extent in 3D cultures. This occurs through a direct activation of E‐cadherin gene promoter by ERα/LKB1‐complex. The impact of E‐cadherin on ERα‐positive breast cancer cell proliferation comes from the evidence that in the presence of E‐cadherin siRNA the proliferative effects of adiponectin is no longer noticeable. Since E‐cadherin connects cell polarity and growth, we investigated if the adiponectin‐enhanced E‐cadherin expression could influence the localization of proteins cooperating in cell polarity, such as LKB1 and Cdc42. Surprisingly, immunofluorescence showed that, in adiponectin‐treated MCF‐7 cells, LKB1 and Cdc42 mostly colocalize in the nucleus, impairing their cytosolic cooperation in maintaining cell polarity. The orthotopic implantation of MCF‐7 cells revealed an enhanced E‐cadherin‐mediated breast cancer growth induced by adiponectin. Moreover, tail vein injection of MCF‐7 cells showed a higher metastatic burden in the lungs of mice receiving adiponectin‐treated cells compared to control. From these findings it emerges that adiponectin treatment enhances E‐cadherin expression, alters cell polarity and stimulates ERα‐positive breast cancer cell growth in vitro and in vivo, sustaining higher distant metastatic burden.


| INTRODUCTION
Clinical investigations reported that low adiponectin concentrations are associated with an increased risk of breast cancer growth exhibiting more aggressive phenotype. 1,2 However, in vitro and in vivo studies have provided contradictory results, describing inhibitory or stimulatory effects of adiponectin on breast tumor growth. 3 Recently, we evidenced that the role of this adipokine in affecting breast cancer growth is tightly dependent on ERα status. [4][5][6] This appears to occur through the transactivation of estrogen receptor α (ERα), recruiting liver kinase B1 (LKB1) as its own coactivator at nuclear level. The latter event interferes with the cytosolic LKB1 activation of 5 0 adenosine monophosphate-activated protein kinase (AMPK), the major regulator of intracellular energy balance. 4 However, LKB1, as the main effector of Adiponectin/AdipoR1 signaling, also plays a crucial role in the maintenance of the apical/basal cell polarity, essential for many cellular processes such as mitosis, morphogenesis and motility. [7][8][9] Moreover, E-cadherin-dependent maturation of cellular adherents junction, driving the proper cellular localization of LKB1/STe20-Related ADaptor (STRAD) complex, is crucially involved in AMPK phosphorylation in polarized epithelial cells. 10 E-cadherin is also an estrogen-responsive gene exhibiting multiple estrogen-responsive element. 11,12 Thus, in the present study it was reasonable to investigate how Adiponectin/AdipoR1, through ERα transactivation, could influence E-cadherin expression. E-cadherin is connected with the activation of multiple proliferative pathways, including the extracellular signal-regulated kinase (ERK) cascade, resulting in a drastic increase in cell proliferation in vitro and in primary tumor growth in vivo, and in such way questioning its role as tumor suppressor. 13 Here, we investigated how adiponectin exposure at the doses tested 1 to 30 μg/mL in ERα-positive breast cancer cells was able to influence the dual role of E-cadherin in modulating breast cancer cell proliferation and cell polarity both involved in breast cancer cell growth, migration and invasiveness. L-glutamine and 1% penicillin/streptomycin (Life Technologies, Milan, Italy). ZR-75-1 were routinely maintained in RPMI supplemented with 10% fetal bovine serum, 1% L-glutamine, 1% penicillin/streptomycin, 1% HEPES, 1% sodium pyruvate and 0.24% glucose. BT-20 cells were cultured in MEM containing 10% fetal bovine serum supplemented with 1% L-glutamine, 1% penicillin/streptomycin, 1% sodium pyruvate and 1% of nonessential-amino acids. Cells were switched to a phenol-red free medium without serum at least 24 hours before each experiment to reduce steroid concentration. The cells, cultured at 37 C in a humidified 5% CO 2 atmosphere, were untreated (Control, C) or treated with globular adiponectin (Prospect, Rehovot, Israel) 1, 5 or 30 μg/mL (A1, A5 and A30, respectively) in a phenol-red free medium containing 1% dextran charcoal-stripped fetal bovine serum. All experiments were performed with mycoplasma-free cells (Applied Biological Materials Inc, Vancouver, Canada).

| Trypan blue cell count assay
Cells (5 Â 10 4 cells/mL) were seeded in 12-well plates. After 24 hours cells were grown in phenol red-free medium containing 1% charcoalstripped Fetal Bovine Serum and untreated or treated with A1, A5 and A30 for 72 hours. Cell numbers were estimated in trypsinized suspensions using Countess II Automated Cell Counter (Invitrogen, Life Technology, Milan, Italy).

| Three-dimensional spheroid culture
MCF-7 and BT-20 cells were plated in polyHEMA-coated 6-well plates as single-cell suspension and untreated or treated with adiponectin for 48 hours. To generate three-dimensional spheroids, the plates were rotated for 4 hours at 37 C. 14 Images were taken at Â10 magnification using a sCMEX-3 microscope camera (Euromex, Spain) and ImageFocus Alpha software. Cell number was determined, after trypsinization of spheroids, by Countess II Automated Cell Counter.

| Plasmids
The plasmid containing the human E-cadherin promoter was given by Dr. Y.S. Chang (Chang-Gung University, Republic of China). 15 The DNA construct encoding the human LKB1 promoter and its 5 0 deleted segments were kindly provided by Dr. G. Singh (McMaster University, Hamilton, Ontario, Canada). 16 2.5 | Real time RT-PCR assay E-cadherin and LKB1 gene expression was evaluated by real-time reverse transcription (RT)-PCR using SYBR Green Universal PCR Master Mix (Thermo Fisher Scientific, Monza, Italy). Each sample was normalized on 18S mRNA content. Relative gene expression levels were calculated as previously described. 17 Primers used are listed in Table S1.

| Immunoblot and immunoprecipitation analysis
For immunoblot analysis, equal amounts of protein extracts were resolved on 8% to 10% SDS-polyacrylamide gels, as described, 18 and probed with primary antibodies: E-cadherin (#sc-8426), LKB1 (#sc-32 245), ERα (#sc-8002), pCAF (#sc-13 124), Cdc42 (#sc-8401) and β-Actin (#sc-69 879), Santa Cruz Biotechnology, Milan, Italy. E-cadherin immunoprecipitation was performed by precleaning 500 μg of total protein lysates for 1 hour with protein A/G-Agarose beads (Santa Cruz, Biotechnology) at 4 C and then centrifuged at 12 000g for 5 minutes. Exactly 10 μL of anti-E-cadherin in HNTG buffer (20 mM HEPES, pH 7,5, 150 mM NaCl, 0.1% Triton X-100, 10% glycerol, 0.1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin) were added to the precleaned protein extracts and incubated overnight at 4 C rotating. The antigen-antibody complex was retrieved by incubation with protein A/G-agarose for 2 hours in HNTG buffer rotating. The beads containing bound proteins were washed three times in Â1 Wash Buffer (100 mM Tris-HCl p H 7.5, 5 M NaCl), then denatured by boiling in Laemmli sample buffer and analyzed by immunoblot to identify the co-precipitating effector proteins. Immunoprecipitation with protein A/G or GFP-nAb Agarose alone was used as the negative control. 19 The antigenantibody complex was detected incubating the membrane with a peroxidase-coupled goat anti-mouse or anti-rabbit IgG and revealed using the ECL system (Santa Cruz Biotechnology, Milan, Italy). Images were acquired by autoradiography or using IBright (ThermoFisher Scientifics, Milan, Italy) and analyzed by ImageJ software ( Figure S1).

| Transfections and luciferase assay
Cells were transfected using the FuGENE 6 reagent (Roche Diagnostics, Monza, Italy), with a mixture containing 0.5 μg of human E-cadherin or LKB1 promoter constructs and 2 ng Renilla luciferasethymidine kinase. After 24 hours, cells were untreated or treated with adiponectin in 1% dextran charcoal-stripped fetal bovine serum for 48 hours. Luciferase activity was measured using a Dual Luciferase kit (Promega, Milan, Italy) as previously reported. 4,20 Firefly luciferase values were normalized to the internal transfection control provided by the Renilla luciferase activity.

| Gene silencing experiments
Cells were transfected with RNA duplex of stealth siRNA (Qiagen, Milan, Italy) targeted for the human ERα, LKB1, or E-cadherin mRNA sequence, using Lipofectamine 2000 (Thermo Fisher Scientific, Monza, Italy). A stealth siRNA (scramble), lacking identity with known gene targets, was used as control for nonsequence-specific effects.
After 6 hours transfection, medium was renewed and silencing efficacy was monitored by immunoblot.
2.10 | DNA affinity precipitation assays (DAPA) Binding of nuclear ERα to ERE DNA motifs was assessed in vitro using nuclear extracts prepared from MCF-7 cells, as previously reported. 21 Briefly, cells were serum starved for 24 hours and then untreated or treated with adiponectin for 1 hour. The assay was carried out in a Oligonucleotide sequences were reported in Table S2. Unlabeled probes were used as negative controls.
The images were taken with FV31S-SW software.

| Wound healing scratch and Golgi reorientation polarity assays
Cells were grown to confluence, wounded and treated with adiponectin. To image Golgi positioning, cells were fixed at 48 hours postwounding and stained with anti-GM-130 antibody. Nuclear staining was performed by DAPI. The percentage of cells with Golgi facing wound was measured as previously described. 22 Images are acquired using a confocal laser scanning microscope (Olympus, Fluoview FV3000, London, UK).

| Proximity ligation assay
Proximity ligation assay (PLA) was performed using Duolink Detection Kit (Merck, Darmstadt, Germany). MCF-7 and BT-20 cells were seeded in the appropriate medium in 8-well chambered slides and grown for 24 hours at 37 C and 5% CO 2 , to allow cell adhesion. After 24 hours, media were replaced with medium supplemented with phenol-red free medium containing 1% charcoalstripped medium and the cells were treated with A1, A5 and A30. and image analysis was performed using ImageJ.  Orthotopic tumor size was measured twice a week by caliper, and tumor volumes (mm 3 ) were calculated using the formula: TV = a Â (b 2 )/2 (a = tumor length and b = tumor width), in millimeters. At fourth week, the animals were euthanized following standard protocols; the tumors were sectioned from the neighboring connective tissue, frozen in nitrogen and stored at À80 C for further analyses.

| Metastatic study
A total of 1 Â 10 6 cells/100 μL were injected into the median tail vein as described. 23 Mice were weighed twice a week and monitored for mobility, respiratory distress and signs of pain. Six weeks after cell injection, mice were sacrificed following standard protocols. Organ examination allowed to identify lungs as destination sites of disseminated tumor cells. Lungs were perfused with PBS, inflated for formalin fixation and paraffin embedding. Lung metastatic lesions were analyzed using hematoxylin and eosin (H&E) and five sections per lung were counted using Imagescope software (Aperio, Leica Biosystems, Milan, Italy). Vector Laboratories, Burlingame, CA). The primary antibody was replaced by normal serum in negative control sections. Stained slides were visualized using Olympus BX41 microscope and the images were taken with CSV1.14 software, using CAM XC-30 for image acquisition.

| Kaplan-Meier analysis
A database for survival analysis was established as described previously. 24 The entire database has 1877 samples from 17 datasets with available overall survival data. The average follow-up is 80.17 months.

Cox proportional hazards regression analysis was performed, and
Kaplan-Meier plots were drawn to visualize survival differences. Samples were grouped into two cohorts based on the gene expression levels, and all possible cutoff values between the lower and upper quartiles of expression were evaluated. False discovery rate was computed to correct for multiple hypothesis testing. ER status was determined using the gene expression data and the analysis was performed separately in ER-positive (n = 1308) and ER-negative (n = 569) patients.

| Statistical analysis
Each data point represents the mean ± SD of at least three independent experiments. In vitro data were analyzed by Student's t test using the Prism 7.0 (GraphPad Software, La Jolla, CA) software program ( Figure S1). Statistical comparisons for in vivo studies were examined by one-way ANOVA with Bonferroni post hoc testing. A value of P < .05 was considered statistically significant.

| Adiponectin/adipoR1 signaling activates E-cadherin gene promoter
To ascertain a direct action of adiponectin on E-cadherin gene expression at transcriptional level, MCF-7 cells were transiently transfected with a luciferase reporter construct containing the human E-cadherin full-length promoter ( Figure 1F). It was extremely intriguing to observe that E-cadherin promoter activity was significantly increased in MCF-7 cells treated with adiponectin ( Figure 1G). The sequence analysis of the E-cadherin promoter showed the presence of two Sp1 sites (À144/À132 and À51/À39; Figure 1F), previously reported to be nonclassic ERα-responsive sites. 11 Figure 1H).
Finally, a crucial role of ERα and LKB1 in adiponectin-modulation of E-cadherin expression raised by the evidence that, in the presence of specific siRNAs, the activation of E-cadherin gene promoter ( Figure 1G) and the upregulation of E-cadherin protein levels ( Figures 1I and S1C,D) were no longer noticeable. Similar results were obtained in ZR-75-1 ERα-positive cells ( Figure S4).

| Adiponectin increases LKB1 expression in ERα-positive breast cancer cells
Adiponectin exerts its effects through the activation of its canonical signaling pathway LKB1/AMPK/mTOR. 26 It is worth to remark that LKB1, acting as ERα coactivator, promotes proliferation in adiponectin-treated breast cancer cells. 4 Here, we demonstrated that adiponectin was able to upregulate LKB1 mRNA and protein contents only in ERα-positive cells (MCF-7 and ZR-75-1) in a dose-related manner (Figures 2A,B and S1E). This promoter ( Figure 2D). This allowed the identification of a sequence (À1626 to À1598) containing three half-ERE putative sites ( Figure 2D), 16 as responsible of the upregulatory effects induced by adiponectin, on LKB1 promoter activity in MCF-7 cells ( Figure 2E).  Figure 2G).
The crucial involvement of ERα in mediating the up-regulatory effect on LKB1 expression upon adiponectin exposure was demonstrated by the evidence that in the presence of either ERα siRNA or the pure antiestrogen ICI 182780, the latter event was no longer evident in MCF-7 cells (Figures 2H and S1G,H).
Thus, we may reasonably hypothesize that upon adiponectin exposure, a new synthetized pool of LKB1 and E-cadherin proteins, mediated by transactivation of ERα, may enrich the endogenous ones.
It is interesting to note that spheroids formation is dramatically

| Modulatory role of adiponectin on cell polarity
Immunofluorescence assay revealed that in adiponectin-treated MCF-7 cells, LKB1 was mostly compartmentalized in the nucleus, while in BT-20 cells LKB1 localized with E-cadherin in the cytosolic compartment ( Figure S6).
Polarity is essential for mammary epithelial integrity, and relies on the capability of LKB1 to regulate a complex network of a big family of GTP-ase proteins. Among them, Cdc42 controls cell polarity through the activation of the downstream serine/ threonine kinase PAK. 28 It has been reported that LKB1 forms a cytosolic complex with active Cdc42 and PAK, regulating cell polarity. 22 Immunoblot analysis revealed that Cdc42 protein level increased in parallel with LKB1 expression in MCF-7 cells treated with adiponectin ( Figure S7A,B). In the presence of specific ERα siRNA, the adiponectin action on Cdc42 expression was still evident, highlighting that it is independent on estrogen signaling ( Figure S7A). In contrast, using the specific LKB1 siRNA, the upregulatory effect seems to be no longer noticeable ( Figure S7B), suggesting the dependence of Cdc42 expression on LKB1 content through a mechanism not yet explored. All this was not reproduced in BT-20 cells ( Figure S7C).
Immunofluorescence assays showed that Cdc42 mostly coloca- cytosolic polarized region was confirmed by Duolink assay in BT-20 cells ( Figure 3B), wherein an intrinsic activity of Cdc42 upon adiponectin treatment was evident ( Figure S8B). In contrast, in adiponectintreated MCF-7 cells both proteins colocalize in the nucleus ( Figure 3B,C) and Cdc42 intrinsic activity was no longer noticeable ( Figure S8B).
To better clarify the role of adiponectin in the regulation of cell polarity, we investigated the orientation of Golgi apparatus, as indicator of cell polarization, in a wound healing assay.
To this end, the GM130, a cis-Golgi marker, was immunostained in MCF-7 and BT-20 cells, followed by a quantitative measure of the Golgi reorientation upon adiponectin treatment ( Figure 4A). 22

| Adiponectin action on breast tumor growth in vivo
To  Figure 6A). Immunohistochemistry analysis revealed that mice injected with adiponectin-treated MCF-7 cells exhibit a higher metastatic burden in the lungs compared with untreated ones, resulting in 1.7-fold increase in the number of metastases for section ( Figure 6B).
On the contrary, no substantial changes in mice injected with adiponectin-treated BT-20 cells were observed ( Figure 6A,B). Thus, in vivo data confirm that adiponectin is able to stimulate tumor growth of ERα-positive cells, exhibiting an altered cell polarity and displaying also a significant higher distant metastatic burden respect to untreated cells. It is worth to remark that LKB1, a key molecule of adiponectin signaling, acts as ERα coactivator. 4,39,40 This emerges from our previous findings demonstrating that adiponectin increased LKB1/ERα protein interaction in the nucleus of MCF-7 cells, favoring the recruitment of LKB1 as ERα coactivator on an ERE sequence in the promoter region of pS2, a classic estrogen-dependent gene. All this has been previously confirmed by the evidence that in the presence of siRNA for LKB1, the latter event was no longer noticeable upon adiponectin exposure in ERα-positive breast cancer cells. 4 The recruitment of LKB1 as coactivator of ERα tethers it in the nucleus, hampering its optimal cytosolic interaction with the scaffold protein STRAD, essential for the activation of AMPK, the master regulator of cellular energy balance. 4 In this concern, the Cancer Genome Atlas (TCGA) breast data set evidenced the deregulation of the expression of one core polarity protein in 65% of luminal A/B, 79% of HER2 and 95% of basal-like breast cancer. 42,43 Our results demonstrated that LKB1 and Cdc42, the two major proteins generally present in cellular polarized region, colocalize with E-cadherin in the cytosol of ERα-negative cells, addressing very likely their cytosolic cooperation in maintaining cell polarity.
Cdc42 is a small GTPase of the Rho family able to control multiple signal transduction pathways. 28