Inhibition of mTORC1 induces loss of E-cadherin through AKT/GSK-3β signaling-mediated upregulation of E-cadherin repressor complexes in non-small cell lung cancer cells

A549 cells were purchased from ATCC (Manassas, VA, USA), and 293FT cells were purchased from Invitrogen (Gaithersburg, MD, USA). H2009, H596, and H1650 cells were obtained from the Korean Cell Line Bank (Seoul, Korea). pLKO.1-Raptor 1 shRNA (plasmid #1857), −Raptor 2 shRNA (plasmid #1858), −Rictor 1 shRNA (plasmid #1853), −Rictor 2 shRNA (plasmid #1854), −TSC2 shRNA (plasmid #15478), psPAX2 (plasmid #12260), and pMD2. G (plasmid #12259) were obtained from Addgene (Cambridge, MA, USA). Antibodies, unless otherwise stated, were obtained from Cell Signaling Technology and are listed in Additional file 1: Table S1 (Danvers, MA, USA).

Clinical information of the 305 patients, who had taken rapalogs under the diagnosis of neoplastic diseases (n = 198) or solid organ transplantation recipient (n = 107) between September 2009 and September 2013, was retrospectively reviewed. Patients’ diagnoses are as follows: renal cell carcinoma (n = 151), thyroid cancer (n = 21), malignant lymphoma (n = 10), neuroendocrine carcinoma (n = 7), hepatocellular carcinoma (n = 5), breast cancer (n = 2), and rectal cancer (n = 2). All the solid organ transplantation recipients had taken rapamycin under the diagnosis of kidney recipient (Table 1). Rapalog-induced interstitial pneumonitis was identified by radiologic diagnosis of chest CT images with exclusion of pulmonary infection or other cause of pulmonary infiltration. Chest CT scan was performed at the onset of respiratory symptoms and signs, not routinely serial follow up exam. Inclusion criteria for drug-induced interstitial pneumonitis of chest CT images were followed by the suggestion of Endo et al. [15]. The use of clinical information received exemption from IRB deliberation.

Lentiviral shRNAs were generated as described elsewhere [16]. Cells transduced with pLKO.1 lentiviral vector were maintained in RPMI medium containing 5% fetal bovine serum. Puromycin (1.0 μg/mL) was added for A549 and H1650 cells, while concentrations of 2.0 μg/mL and 3.0 μg/mL were used for H460 and H1299 cells, respectively.

Cells were harvested using 2X LSB lysis buffer containing protease inhibitor cocktail and phosphatase inhibitor cocktail (Sigma, St. Louis, MO, USA) on ice. After sonication, BCA protein assay reagent (Thermo Scientific, Rockford, IL, USA) was used for protein quantification. Protein lysates (30–50 μg) were separated by gel electrophoresis on 7.5% to 12% polyacrylamide gels and analyzed by western blot using nitrocellulose membranes (Bio-Rad Laboratories, Inc., Richmond, CA, USA). Expression level of each protein was measured using ImageJ (http://​imagej.​nih.​gov/​ij/​) and quantified relative to that of β-actin.

5 × 10⁵ cells were plated in 6-well plates containing a sterilized coverslip. The next day, cells were fixed with 4% formaldehyde in PBS, incubated in blocking solution containing 5% BSA in PBS, and then anti-E-cadherin mouse (1:100) and anti-Vimentin rabbit antibodies (1:200) were added for 16 hr. The next day, cells were washed and anti-mouse IgG (Alexa Fluor 555 conjugate) and anti-rabbit IgG (Alexa Flour 488 conjugate, Cell Signaling Technology) secondary antibodies were added. Nuclei were counterstained with DAPI (1:1000, Sigma) in PBS and imaged using an LMS510 confocal microscope (Carl Zeiss, Oberkochen, Germany).

Total RNA was extracted using TRI Reagent® (Ambion, Austin, TX, USA). Reverse transcription and quantitative PCR were performed using High Capacity cDNA reverse transcriptase and TaqMan® FastUniversal PCR Master Mix (2X), respectively. qRT-PCR analysis was conducted using TaqMan® gene expression assay reagents, and the StepOnePlus Real-Time PCR system (Applied Biosystems, Carlsbad, CA, USA) with an inventoried primer-probe set described in Additional file 2: Table S2. IPO8 was amplified in a primer-limited fashion using primers labelled with VIC dye as an endogenous control.

During the recent 4 years, 16 (5.2%) out of 305 patients, who had taken rapalogs under the diagnosis of either neoplastic diseases or solid organ recipient, experienced clinically apparent interstitial lung disease in the study institutes (Figure 1A and Table 1). Among them, 7 patients showed severe interstitial pneumonitis (grade ≥ 3). The mean time to development of interstitial pneumonitis was 6.4 months (5.8 months for everolimus, 4.6 for temsirolimus, and 8.6 for rapamycin). Together with the loss of E-cadherin expression in the LAM, development of interstitial pneumonitis during rapalogs treatment brought us a question whether aberration mTOR signaling impacts the expression of E-cadherin and EMT.

To explore this, A549 cells were treated IGF-1 and expression of E-cadherin was evaluated by immunoblotting. Treatment with IGF-1 decreased E-cadherin expression in time- and dose-dependent experiments, and E-cadherin expression was more dependent on time, reaching a nadir at 48 hr after IGF-1 treatment (Figure 1B and C). There was a dose-dependent decrease in E-cadherin mRNA with IGF-1 treatment, suggesting that the loss of E-cadherin occurred at the transcriptional level (Figure 1D). Next, NSCLC cells were treated with rapamycin, an mTOR inhibitor, and expression of E-cadherin was evaluated by immunoblotting. Treatment with rapamycin clearly increased E-cadherin protein and mRNA expression (Figure 1E,F and G). We also examined the effect rapamycin on the expression of other components of the adherens junction complex by immunoblotting. Inhibition of the mTOR pathway with rapamycin increased expression of E-cadherin, β-catenin, catenin-δ-1, and α-E-catenin (Figure 1G). Taken together, these findings suggest that the mTOR pathway is involved in regulating the expression of E-cadherin and the other components of adherens junction complex.

Since recent reports indicate that mTOR functions as a part of mTORC1 or mTORC2 depending on its binding partners, and each complex has different inherent roles, we questioned whether loss of E-cadherin is mediated through mTORC1 or mTORC2. To investigate this, A549 cells were transduced with shRNA against Raptor, Rictor, or TSC2, which regulate mTORC1, mTORC2 and Rheb, respectively. Samples were then blotted for E-cadherin. Raptor shRNA-transduced cells lost E-cadherin expression, while Rictor and TSC2 shRNA-transduced cells did not (Figure 2A). To further confirm this finding, different Raptor shRNA constructs were transduced into A549 and H2009 NSCLC cells and the expression of E-cadherin was evaluated by immunoblotting and real-time PCR. E-cadherin expression decreased in A549 and H2009 cells when they were transduced with either Raptor shRNA construct, and the loss was more prominent in A549 cells (Figure 2B). E-cadherin mRNA expression also decreased in A549 and H2009 cells transduced Raptor shRNA (Figure 2C). These findings suggest that inhibiting mTORC1 by silencing Raptor reduces E-cadherin expression at the transcriptional level.

Because feedback regulation of pAkt is one of the functions of mTORC1, we next evaluated the phosphorylation of Akt at Ser473 in Raptor-silenced NSCLC cells. Increased phosphorylation of Akt at Ser473 was observed in A549, H2009, H460, and H1299 Raptor-silenced NSCLC cells, causing inhibitory phosphorylation of the Akt substrate GSK-3β at Ser9 (Figure 3). To further confirm hyperactivation of Akt/GSK-3β signaling in Raptor-silenced NSCLC cells, siRNA against Raptor was introduced into A549, H460, H1299, and H2009 NSCLC cells (data not shown). All NSCLC cells treated with Raptor siRNA showed increased levels of pAKT-Ser473 and pGSK-Ser9 compared to those transduced with scrambled siRNA. These findings suggest that Akt/GSK-3β pathway aberrations are a general finding in cells with mTORC1 disrupted by Raptor silencing and that Akt phosphorylation leads to functional activation of Akt/GSK-3β signaling.

Because transcriptional inhibition of E-cadherin is frequently mediated by E-cadherin repressor complexes and we observed decreased E-cadherin mRNA expression in Raptor-silenced NSCLC cells, we evaluated E-cadherin repressor complex mRNA expression by qRT-PCR. Among the E-cadherin repressor complexes tested, Snail, Zeb2, and Twist1 had increased expression in Raptor-silenced A549 cells and Zeb2 and Twist1 were elevated in Raptor-silenced H2009 cells (Figure 4A). We postulated that inhibitory phosphorylation of GSK-3β at Ser9 by Akt might cause the increase in Snail, Zeb2, and Twist mRNA. To test this, A549 parental cells were treated with LiCl, a GSK-3β inhibitor, and E-cadherin repressor complex mRNA expression was evaluated by qRT-PCR. Indeed Snail, Zeb2, and Twist1 mRNAs were elevated in cells treated with LiCl (Figure 4B). Because Snail is mainly regulated by β-TrCP mediated proteosomal degradation secondary to GSK-3β-mediated phosphorylation [17], we were curious about the contribution of the mRNA level elevation to the expression of Snail protein. LiCl treatment slightly increased the expression of Snail and its effect was comparable to that of IGF-1. Combined treatment with LiCl and MG132 strongly increased expression of Snail. Taken together, these findings suggest involvement AKT/GSK-3β signaling in the loss of E-cadherin repressor complexes in Raptor-silenced cells (Figure 4C).

EMT entails morphological and phenotypic changes of epithelial cells leading to loss of cellular polarity and intercellular adhesion, and the development of migratory and invasive properties. Phase-contrast imaging of Raptor-silenced cells shows a transition from the typical cobblestone appearance to elongated spindle shapes. These findings are more prominent in A549 cells than in H2009 cells. To further elucidate the EMT in Raptor-silenced cells, immunocytochemical staining against E-Cadherin and Vimentin was performed. Raptor-silenced A549 cells showed a marked increase in Vimentin and decrease in E-cadherin expression whereas H2009 cells did not exhibit changes in the expression of E-cadherin and Vimentin (Figure 5A). Then mesenchymal markers, α-smooth muscle actin (SMA), Vimentin, and N-cadherin, were blotted in scrambled- and Raptor-shRNA-transduced cells. Similar to the findings in the morphologic study, Raptor-silenced H2009 cells did not show changes in the expression of Vimentin, N-cadherin, and SMA. Expression of Vimentin and N-cadherin was increased in Raptor-silenced A549 cells while that of SMA was unchanged, suggesting that disruption of mTORC1 leads to EMT in a subset of cells (Figure 5B).