Inverse agonism at the Na/K-ATPase receptor reverses EMT in prostate cancer cells

Abstract

The surface expression of Na/K-ATPase α1 (NKA) is significantly reduced in primary prostate tumors and further decreased in bone metastatic lesions. Here, we show that the loss of cell surface expression of NKA induces epithelial mesenchymal transition (EMT) and promotes metastatic potential and tumor growth of prostate cancer (PCa) by decreasing the expression of E-cadherin and increasing c-Myc expression via the activation of Src/FAK pathways. Mechanistically, reduced surface expression of NKA in PCa is due to increased endocytosis through the activation of receptor NKA/Src complex. Using a high throughput NKA ligand screening platform, we have discovered MB5 as an inverse agonist of the NKA/Src receptor complex, capable of blocking the endocytosis of NKA. MB5 treatment increased NKA expression and E-cadherin in PCa, which reversed EMT and consequently decreased the invasion and growth of spheroid models and tumor xenografts. Thus, we have identified a hitherto unrecognized mechanism that regulates EMT and invasiveness of PCa, and demonstrated for the first time the feasibility of identifying inverse agonists of receptor NKA/Src complex and their potential utility as anti-cancer drugs. We, therefore, conclude that cell surface expression of α1 NKA can be targeted for the development of new therapeutics against aggressive PCa, and that MB5 may serve as a prototype for drug development against EMT in metastatic PCa.

INTRODUCTION

Prostate cancer (PCa) is the second most common type of cancer in males and is treatable if detected at early stages with a five-year survival rate of nearly 100%. However, at advanced stages when the cancer spreads to distant organs through metastasis, this survival rate drops to only about 27% [1,2]. Androgen deprivation therapy (ADT) is the first line of therapy for advanced PCa, but a significant fraction of these tumors become resistant and progress to metastatic castration resistant state (CRPC) for which there is no effective cure. Because most carcinomas are of epithelial origin, tumor cells reactivate a developmental program called Epithelial-Mesenchymal Transition (EMT). Tumor cells undergoing EMT alter their apical-basal polarity and lose their adherens junctions by activating mesenchymal genes, which in turn transcriptionally repress cell adhesion molecules. This enables them to escape the stressful tumor microenvironment by assuming a motile fibroblast-like phenotype to invade the vasculature, and subsequently colonize at distant sites. The major hallmarks of EMT are a loss of adhesion junction and cell polarity proteins (E-cadherin, ZO-1/2, claudins and occludin) and gain of expression of mesenchymal genes (e.g. SNAIL, ZEB1/2, TWIST) [3,4]. Novel molecular targets and therapeutics focused on EMT are an attractive approach in the treatment of metastatic PCa, which accounts for nearly 90% of PCa patient deaths.

Na/K-ATPase α1 (NKA), a transmembrane ion pump [5] and a fundamental signaling mechanism in cell proliferation and differentiation ([6–8] [9]), may be one such target. Indeed, NKA expression at the plasma membrane is an important determinant of epithelial apical-basal polarity and maintenance of cell-cell adhesion junctions [10,11], a feature that is frequently lost in EMT. Consistently, reduced NKA subunit (α and β) expression has been reported in association with EMT in both cell and animal models of fibrosis and carcinoma [12–14]. Highly relevant in the context of the present study, we recently reported that NKA expression levels are inversely correlated with metastatic spread of prostate carcinomas. Genetically targeted loss of α1 NKA in PCa cells subsequently causes a metabolic switch from oxidative phosphorylation to aerobic glycolysis (Warburg effect) through Src kinase activation, and increased tumor volume in a mouse xenograft model [15]. Clinically, NKA α1 expression is largely undetectable in bone metastatic lesions of PCa patients [15], indicative of translational potential of increasing NKA expression as a therapeutic approach. Collectively, these studies have therefore established a strong and clinically significant link between NKA expression and invasiveness/metastasis of prostate carcinoma. They also suggest that EMT secondary to decreased NKA expression could be targeted to decrease metastasis/invasiveness of prostate carcinoma.

Alteration of NKA cellular distribution in PCa cells is secondary to an increase of α1 NKA receptor endocytosis [16]. This mechanism is NKA-specific and can be modified pharmacologically by modulating its receptor function. Cardiotonic steroids (CTS) are the archetypal and best-studied NKA ligands. They bind to and inhibit the enzymatic activity of NKA by stabilizing the protein in its E2P conformation [17]. Because E2P represents an active conformation for Src and α1 NKA interaction [18–22], CTS such as ouabain are agonists of the receptor α1 NKA/Src complex. Accordingly, these compounds stimulate protein and lipid kinases, increase Reactive Oxygen Species (ROS) production and induce the endocytosis of α1 NKA [23–25]. We have developed a high throughput screening platform to identify novel non-CTS α1 NKA ligands and assess their molecular actions on the signaling function of α1 NKA as either agonists or inhibitors. Using this platform, we have identified a group of small molecules with xanthone backbone that bind NKA but do not provoke NKA-mediated signal transduction [26]. Hence, we surmised that this family of compounds could be modulators of NKA cell surface expression and inhibitors of EMT in metastatic PCa cells.

Using a combination of NKA gene targeting, EMT markers analyses, pharmacological characterization, and functional assays in three dimensional (3D) spheroids and tumor xenograft models, the studies presented herein suggest that MB5, a small molecule inverse agonist of NKA receptor function, has the potential to block metastasis and reduce tumor growth by reversing EMT in PCa.

RESULTS

Knock-down of α1 NKA induces EMT and promotes PCa cells migration and invasion.

To experimentally assess the impact of loss of NKA expression on tumor growth and metastatic potential, we xenografted DU145 and DU145-derived NKA KD PCa cells (~50% reduced by RNAi) [15] into NOD/SCID mice to generate tumors (Fig. 1a). All tumors grew locally at the injection sites except one tumor from KD cells, which metastasized to the bones. Cell lysates were prepared from one half of each tumor sample, whereas tumor cells were isolated from the other half by enzymatic digestion. Fig. S1a shows total α1 NKA protein expression in each tumor sample. Subclone 5 (isolated from a DU145 xenograft) as well as subclones 4 and 2 (isolated from KD xenografts) were selected for further comparison. Western blot analyses (Fig.1b) showed that α1 NKA expression was significantly reduced in the subclones compared to the parental cell lines. Subclone 2, a cell line isolated from the only bone metastatic tumor, exhibited the lowest expression of α1 NKA (~80% reduced expression compared to KD cells). Expression of β1 NKA, which also has a tumor suppressor function, was only modestly reduced in subclones 4 and 2 compared to subclone 5 (Fig. 1c). Reduction in α1 NKA expression was further verified by immunostaining subclones 5 and 2 with a monoclonal anti-α1 NKA antibody (Fig. S1b). Consistent with our previous findings [15,16,27], decrease in α1 NKA expression was inversely associated with activation of Src kinase and its effector protein FAK (phosphoprotein/total protein ratios) and Myc expression (Fig. 1d). This was accompanied by a modest increase in expression of cell cycle proteins cyclin D1, E1 and the cell proliferation marker, PCNA (Fig. S1c).

Fig. 1. Loss of α1 NKA in DU145 induces EMT and promotes invasion.

(A) Generation cell subclones from DU145 and α1 NKA knockdown DU145 (KD) cells. Representative immunoblots for α1 and β1 NKA expression are shown for comparison. (B) Quantitative analysis of α1 NKA expression (Western blot analysis), relative to tubulin. **p<0.01, ***p<0.001 as indicated (N=4, one-way ANOVA). (C) Quantitative analysis of β1 NKA expression relative to tubulin. *p<0.05, n.s=not significant, as indicated (N=4, one-way ANOVA). (D) Representative immunoblots for basal phospho and/or total forms of Src, FAK and Myc. Tubulin blot confirms equal loading. *p<0.05, **p<0.01 and ***p<0.001 relative to clone 5 (N=3–4, one-way ANOVA). (E) Representative phase-contrast images of subclones 5, 4 and 2. (N=4, scale bar = 50μm). (F) Representative immunoblots for epithelial (E-cadherin, ZO-1, ZO-2, occludin) and mesenchymal markers (SNAIL and ZEB1) in indicated subclones. (G) Quantitative analyses (Western blot analysis)*p<0.05 and **p<0.01 relative to clone 5 (N=4, one-way ANOVA). (H) qPCR analyses of EMT markers. *p<0.05 and **p<0.01 compared with sub-clone 5 (N=6, one-way ANOVA). (I) Relative cell migration at 16 hours (Boyden chamber assay) **p<0.01 compared with sub-clone 5 (N=6, one-way ANOVA). (J) Spheroid formation assay at day 7 (representative phase contrast images), (N=4, scale bar = 50μm). (K) Spheroid invasion assay. Representative images at indicated timepoints. (N=8, scale bar = 50μm) (L) Quantitative analysis of invasion. ***p<0.0001(N=8, Students t test). (M) Representative immunoblot for MMPs secreted by spheroids. Ponceau stained nitrocellulose membrane is shown as loading control (N=3). ANOVA, analysis of variance; EMT, epithelial-mesenchymal transition; MMP, matrix metalloproteinase; NKA, Na/K-ATPase α1

In 2D cultures, severely decreased expression of α1 NKA was associated with a change in phenotype from epithelial to mesenchymal (fibroblastic) morphology (Fig.1e). This effect was most pronounced in subclone 2, which expressed the least amount of α1 NKA and exhibited an extreme spindle shape and loss of cell-cell attachment. Subclone 5, with the highest NKA expression, retained the typical cobblestone epithelial phenotype of DU145 cells, whereas subclone 4 displayed an intermediate phenotype. Further Western blot analyses of EMT signature markers (Fig. 1f and g) revealed that several epithelial markers like E-cadherin, β-catenin, ZO1/2 and occludin were significantly downregulated in subclones 4 and 2. This was accompanied by a significant increase in mesenchymal proteins SNAIL and ZEB1, thus, consistent with an EMT phenotype. This EMT phenotype was further verified by qPCR of EMT markers such as E-cadherin, vimentin, or N-cadherin (Fig. 1h). Because a gain of EMT phenotype is associated with increased migratory capability, we conducted a Boyden chamber assay (Fig. 1i), which revealed that subclone 2 cells migrated significantly faster than both subclones 5 and 4.

EMT is associated with loosened cell-cell contact allowing metastatic dispersion of cancer cells, therefore we tested if loss of α1 NKA contributes to decreased cell-cell adhesion and increased invasion. In 3D cultures [28] (Fig. 1j), subclone 5 cells formed compact homogenous spheroids after 7 days, indicating their capability to form strong intracellular adhesion. In contrast, subclones 4 and 2 formed loose grape-like stellate spheroids and also invaded into the matrix during this time, consistent with their inability to maintain intracellular adhesion under reduced α1 NKA expression. The invasive capability of the subclones was next tested using a spheroid invasion assay. As illustrated in Fig. 1k, compact spheroids were generated by the hanging drop technique and embedded in a 3D matrix. Cells from subclone 2 spheroids invaded the matrix as early as 12 h and formed extensive invasive structures by 72 hours. In contrast, subclone 5 spheroids did not invade into the matrix even after 72h of culture (Fig 1k, l). To complement these findings, matrix metalloproteinase (MMP) secretion into 3D culture media was measured as an indicator of PCa cells’ ability to degrade extracellular matrix and enable metastasis [29]. Subclone 2 secreted significantly more MMP2 and MMP9 into the culture media than subclone 5, as measured by Western blot analyses of conditioned media (Fig. 1m).

The above cell KD models were derived from the DU145 cell model of CRPC. It was therefore important to verify the proposed concept in another PCa cell model. Accordingly, we knocked down α1 NKA expression in the C4–2 cell CRPC model, using RNA interference (RNAi). Although the knockdown efficiency was only 40% (Supplementary Fig. 1e), it resulted in an EMT phenotype as evident from the morphological change from the tightly clustered epithelial colony formed by parental C4–2 cells to the significant loss of cell-cell attachment and fibroblastic morphology of α1 NKA KD cells (Supplementary Fig.1d). This was accompanied by significant reduction in epithelial markers such as occludin, E-cadherin and ZO-1 expression and upregulation of mesenchymal markers in the KD cells (Fig. s1e and f). In agreement with a Src kinase regulatory role for α1 NKA [18], Src kinase and its effector proteins FAK and ERK were significantly activated in the KD cells (Fig. s1g). The spheroid invasion assay revealed that reduced expression of α1 NKA enabled KD cells to invade into the matrix in 2 days, which is in sharp contrast to parental C4–2 spheroids that exhibited no invasion even after 7 days in culture (Fig. s1h).

α1 NKA rate of endocytosis correlates with EMT in PCa cell lines.

Our previous studies have shown that NKA expression in PCa cells is regulated by a posttranslational mechanism (increased endocytosis) rather than a transcriptional mechanism, which is supported by findings from other laboratories [12,16]. In the present study, biotinylation assays showed that the rate of endocytosis of α1 NKA was indeed highest in the aggressive subclone 2 compared to the parental DU145 by about 10 fold (Fig. 2a). This increase in endocytosis was inversely proportional to their total α1 NKA expression (Fig. 1b). To determine whether changes in α1 NKA expression correlate with an EMT phenotype, we analyzed α1 NKA expression in four PCa cell lines by Western blot and immunofluorescence. As shown in Fig. 2c, DU145 and PC3 cell lines derived from distant metastatic sites express a significantly lesser amount of α1 NKA than LNCaP, a lymph node metastatic cell line, or its derivative C4–2. Moreover, immunofluorescence analyses (Fig. 2b) revealed that whereas all α1 NKA signal was localized to the plasma membrane in C4–2 cells, a significant amount of α1 NKA resided in intracellular compartments in DU145 and almost all of the signal was observed in the cytosol for PC3 cells. This was associated with an upregulated EMT phenotype in DU145 and PC3 as compared with C4–2 (Fig. 2c, d), thus further supporting our contention that α1 NKA rate of endocytosis correlates with EMT in PCa cells.

Fig. 2. α1 NKA endocytosis and EMT.

(A) α1 NKA endocytosis (internalization) using cell surface biotinylation assay. **p<0.01 compared with DU145 (N=3, one-way ANOVA). (B) Representative confocal images of α1 NKA cellular distribution. DU145 and PC3 imaged at 3X exposure to C4–2(scale bar=50μm). (C, D) Representative immunoblots showing α1 NKA, epithelial marker(left) and mesenchymal marker (right) expression in common PCa cell lines (top). Quantitative analyses (bottom) *p<0.05 and **p<0.01 relative to C4–2 (N=4, two-way ANOVA). ANOVA, analysis of variance; EMT, epithelial-mesenchymal transition; MMP, matrix metalloproteinase; α1 NKA, Na/K-ATPase α1

Overexpression of α1 NKA is an approach to counter tumor growth in the NOD/SCID mouse model.

Since the loss of α1 NKA expression contributes to tumor progression, we next tested whether genetic rescue of its expression counters tumor growth in a xenograft mouse model. Specifically, we rescued the KD cells derived from DU145 with a rat α1 NKA construct-containing expression vector. Rat α1 NKA expression was verified by Western blot analyses (Fig. 3a). In addition, taking advantage of the well-known low affinity of rodent NKA α1 to ouabain compared to human, successful rescue was functionally confirmed by acquired resistance to ouabain-induced cell death using a MTT assay (Fig. 3b). As shown in Fig. 3c, α1 KD cells had increased cell proliferation rate in comparison with parental DU145 cells, but rat α1 rescue decreased the cell proliferation rate to a level similar to DU145. This anti-proliferative effect was correlated with decreased Src activation, Myc expression and total protein tyrosine phosphorylation in α1 NKA rescued cell line in comparison with the KD cells (Fig. 3d, e).

Fig. 3. Genetic rescue of α1 NKA counters tumor growth.

(A) Representative immunoblot showing rat α1 NKA expression (anti-NASE antibody recognizes only rat α1 polypeptide) in rescued cells and parental DU145. Bottom panel shows total α1 NKA expression (α6f antibody recognizes both human and rat NKA α1 polypeptide), (N=3). (B) MTT assay showing effect of ouabain on cell viability of DU145, KD and rat α1 rescued cells. (N=5–6). (C) Cell proliferation assay, *p<0.05 as indicated (N=6, one-way ANOVA). (D) Representative immunoblots showing Src activation (phospho-protein vs. total protein), total phosphotyrosine, Myc and tubulin expression (N=4). (E) Quantitative analysis (Western) *p<0.05 and **p<0.01 relative to DU145, (N=3, one-way ANOVA) (F) Tumor weight from α1 KD and α1 rescued cell xenograft. *p<0.05 (N=10, Students t test). ANOVA, analysis of variance; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; α1 NKA, Na/K-ATPase α1

Finally, when xenografted into NOD/SCID mouse model, rat α1 rescued cells formed significantly smaller tumors than those from KD cells (Fig. 3f). This data set indicated that rescue of α1 NKA expression could represent a novel mechanism for preventing PCa progression. On this basis, we implemented a cell-based assay to identify pharmacological agents that can inhibit α1 NKA endocytosis.

Identification of MB5, an inverse agonist that targets α1 NKA /Src signaling.

As archetypal agonists of NKA receptor function, ouabain and other CTS stimulate cellular signaling by activating α1 NKA/Src signalosome complex, which results in its endocytosis [23]. We reasoned that compounds that inhibit CTS induced signaling could function as inverse agonists of this signalosome complex and prevent α1 NKA endocytosis. We, therefore, turned our attention to a group of hydroxyxanthones previously identified as a new class of Na/K-ATPase ligands using a high throughput in vitro screening platform and a library of 2600 structurally diverse chemicals[26]. Structure-Activity Relationship (SAR) studies of these new ligands revealed a range of pharmacological potency on NKA enzymatic activity, with at least one compound without detectable CTS-like activating (agonist) effect on NKA receptor signaling [26]. Further pharmacological characterization of this small molecule, MB5 (Fig. s2a), revealed a biphasic NKA inhibition curve with a high affinity component (nM), which represented only 20% of Na/K-ATPase activity, and a low affinity component with an IC50 of 10μM (Fig. s2b).

Characterization of MB5 activity on the NKA receptor function known as the NKA/Src binary receptor mechanism [30] was conducted in a pig kidney epithelial cells (LLC-PK1)-based platform using immunostaining and Western blot analyses. The binary NKA/Src receptor model summarized in Fig 4 has been shown to regulate a series of downstream signaling events that include ERK phosphorylation and endocytosis in LLC-PK1 cells and a number of other cells [23,31,32]. According to this model, under normal basal conditions, most NKA/Src binary receptors adopt a conformation whereby Src binds to NKA through two defined sites of interaction and is kept inactive. The change of NKA conformation that occurs upon binding of agonists such as ouabain, results in the release and activation of the Src kinase domain from the “Naktide” site of NKA CD3, while the other interaction (constitutive) persists between NKA CD2 and the SH2 domain of Src [18,30]. As presented in Fig. 5a–c, MB5 (10–100 nM) did not activate NKA receptor function under baseline condition, but dose-dependently inhibited Src and ERK activation induced by the prototypic NKA agonist ouabain. Accordingly, MB5 was also a potent inverse agonist of ouabain-induced α1 NKA endocytosis as confirmed via confocal microscopy using a cell line expressing YFP tagged-murine α1 NKA (Fig. 5d). This effect was specific to NKA-mediated signaling, as MB5 failed to inhibit dopamine- and EGF-induced ERK activation (Fig. s2c–f). In LLC-PK1-derived cells with low NKA receptor constitutive activity due to substantial reduction (over 80%) of α1 NKA expression by siRNA (PY17 cells), increased basal Src and ERK activity is observed due to the abnormally high amount of Src kinase that remains in an active conformation [27]. Increased basal Src and ERK activity is also observed in LLC-PK1-derived cells with reduced NKA receptor constitutive activity that result from mutations in either domain of interaction with Src without alteration of the ion pumping activity (e.g. A425P on the Naktide sequence, or Y260A on the CD2 domain) [15,18,19]. As shown in Fig. 5e, MB5 inhibited ERK phosphorylation in a dose-dependent manner in PY17. On the basis of inverse agonism observed in the presence of ouabain, MB5’s effect in PY17 is best explained by a stabilization of the remaining NKA receptors in an inactive conformation, which in turn normalized basal phospho ERK in those cells. In contrast, MB5 did not reduce high levels of ERK phosphorylation in A425P and Y260A (Fig. s2g), suggesting that MB5 specifically targets the NKA/Src signaling branch of NKA receptor function. Critically, as in PY17, MB5’s inverse antagonism also applied in PCa cells with increased Src kinase activity due to low NKA levels, where MB5 treatment abolished α1 NKA endocytosis in subclone 2 cells in a dose-dependent manner (Fig. 5f).

Fig. 4. Schematic diagram of molecular mechanism.

(A) Effect of tumor microenvironment on α1 NKA/Src receptor complex activation and its endocytosis leading to EMT (B) MB5 treatment blocks α1 NKA/Src receptor complex in inactive conformation and reverses EMT by stabilizing cell-cell attachment. EMT, epithelial-mesenchymal transition; α1 NKA, Na/K-ATPase α1

Fig 5. Validation of MB5 as an inverse agonist of α1 NKA/Src signaling.

(A) Confocal images of ouabain and MB5 on phosphor Src content in LLC-Pk1(N=4). (B) Confocal images of ERK activation/phosphorylation in LLC-PK1 cells, *p<0.05 and **p<0.01, n.s= not statistically significant (N=3, one-way ANOVA). Images are at same scale. (C) Effect of MB5 and ouabain treatment on ERK activation in LLC-PK1 cells (Western blot analysis). **p<0.01 compared with control and ##p<0.01 compared with only ouabain-treated group (N=3, one-way ANOVA). (D) Effect of MB5 on ouabain-induced rat α1 NKA (YFP-tagged) endocytosis(immunostaining). ###p<0.001 compared with control;**p<0.01 and *p<0.05 compared with only ouabain treated cells. (N=3, one-way ANOVA). (E) Effect of MB5 on basal ERK phosphorylation in PY17 cells, *p<0.05 and **p<0.01 compared with control, (N=3, one-way ANOVA). (F) Effect of MB5 treatment on α1 NKA endocytosis ( biotinylation assay) in subclone 2. **p<0.01 and *p<0.05 compared with control (N=3, One-way ANOVA). ANOVA= analysis of variance; α1 NKA, Na/K-ATPase α1

MB5 reverses EMT in PCa cells and reduces their metastatic potential.

Next, we tested whether MB5 as an inverse agonist of NKA receptor function could reverse EMT and stop invasion of PCa cells by rescuing α1 NKA expression in PCa. Subclone 5 and 2 cells were exposed to a long term treatment (0, 24, 48 and 72h) with 100nM MB5. A significant time-dependent increase in α1 NKA and E-cadherin expression, along with a decreased expression of mesenchymal markers SNAIL and ZEB1 was observed, with maximal inhibition occurring at 72 h (Fig. 6a). This was associated with a significant decrease in Src, FAK activation and Myc expression. Treatment of subclone 2 cells with the Src kinase inhibitor PP2 also resulted in increase in E-cadherin expression (Fig. s3a), suggesting that Src kinase activation might be partially responsible for E-cadherin loss through proteasomal cleavage[33] in addition to transcriptional downregulation as observed in Fig.1h. Moreover, expression of the cell proliferation marker PCNA, was either abolished (in subclone 5) or significantly reduced (in subclone 2) by 72 h, indicating the potential for MB5 to reduce cancer cell growth. Increase in E-cadherin and occludin expression by MB5 was further confirmed by immunostaining in subclone 2 cells treated with 1μM MB5 for 24 h (Fig. 6b).

Fig. 6. MB5 inhibited spheroid growth and invasion by reversing EMT and Src/FAK signaling.

(A) Representative immunoblots showing effect of 100nM MB5 treatment on expression of EMT markers, Src/FAK activation and cell proliferation markers (Myc and PCNA) in DU145-derived subclones. Loading control- tubulin, *p<0.05 and **p<0.01 as indicated, (Students t test, N=3–4). (B) Confocal images showing effect of 1μM MB5 treatment on E-cadherin and Occludin expression in subclone 2 after 16 h (N=3, scale bar=50μm). (C)Representative images (top) and quantitative analyses (bottom) showing effect of MB5 treatment on invasion and growth of subclone 2 spheroids. **p<0.01 compared with 0nM (N=4, one-way ANOVA). (D) Representative immunoblots showing effect of MB5 treatment on MMPs secretion by subclone 2 spheroids. N=3, loading control= Ponceau stained nitrocellulose membrane. (E) Boyden chamber migration assay of subclone 2 cells pretreated with MB5, PP2 (Src inhibitor) or FAK inhibitor. DMSO= vehicle, n.s.= not significant, **p<0.01 compared with –DMSO treated group (N=6, One-way ANOVA). (F) Effect of MB5 treatment on growth of sub-clone 5 spheroids, representative images (top) and quantitative analyses (bottom). *p<0.05 and **p<0.01 compared with 0nM (N=4, One-way ANOVA). ANOVA= analysis of variance; DMSO, dimethyl sulfoxide; EMT, epithelial-mesenchymal transition; MMP, matrix metalloproteinase

Functionally, MB5 suppressed spheroid invasion by subclone 2 cells in a dose-dependent manner and significantly reduced spheroid size by 72 h (Fig. 6c). This anti-invasive effect was further verified by Western blot analyses of media (Fig. 6d), which showed that MB5 treatment significantly reduced MMP2 and 9 secretions by subclone 2 spheroids. Finally, Boyden chamber assay (Fig. 6e) showed that pretreatment with 100nM MB5 for 24 h was sufficient to reduce cell migration by 50%, an effect that was comparable to Src or FAK inhibitor treatment. MB5 also significantly reduced the size of subclone 5 spheroids in 3D culture (Fig. 6f) in agreement with its inhibitory effect on PCNA. Taken together, the above studies indicate a molecular mechanism whereby increased surface expression of α1 NKA reigns in Src kinase activity in cancer cells, thus rescuing cell adhesion proteins like E-cadherin from proteasomal degradation and gradually suppressing the EMT phenotype.

Because subclones 2 and 5 were selected and derived in our laboratory as proof-of-concept models based on metastatic potentials and NKA expression levels, the universality of proposed MB5 mechanism and efficacy on NKA signaling, EMT, and invasiveness of PCa cells was further tested independently in PC3 and C4–2 cells. In 3D cultures, vehicle-treated PC3 cells formed loose heterogeneous spheroids, which invaded the matrix significantly by day 7. MB5 treatment suppressed spheroid invasion, and also reduced spheroid size in a concentration-dependent manner (about 25% inhibition at 0.1μM and 50% at 1μM; Fig. 7a). In ultralow attachment plates, control PC3 cells did not form cellular aggregates due to a null mutation of α-catenin gene, in accordance with previous reports [34]. Nonetheless, MB5 treatment increased cell aggregation in a concentration-dependent manner and also significantly increased E-cadherin expression in those cells (Fig. 7b), confirming that MB5 works by tightening cell-cell attachment. Finally, Western blot and cell fractionation analyses (Fig 7c, d) confirmed that 72–120 h of MB5 treatment in PC3 cells significant increased α1 NKA expression, decreased Src activation, and reversed the EMT phenotype. MB5 treatment was also able to upregulate occludin and E-cadherin expression in KD cells from C4–2, while inhibiting the expression of SNAIL, SLUG and Myc (Fig 7e) along with Src activation (Fig. s3b). On the contrary, MB5 treatment at concentrations as high as 2μM (Fig. s3c) did not significantly affect spheroid growth of the parental C4–2 cells, consistent with low α1 NKA endocytosis in these cells (Fig. 2b).

Fig. 7. Effect of MB5 on PC3 and C4–2.

(A) Effect of MB5 treatment on PC3 spheroid invasion and growth. Representative images (top) and quantitative analyses (bottom). *p<0.05 and **p<0.01 compared with 0nM (N=6, one-way ANOVA). Spheroid growth from day 1 to 7, **p<0.01 compared with 0nM (N=5, one-way ANOVA). (B) Representative images of E-cadherin immunostaining in PC3 cells in ultralow attachment plate. (N=4, scale bar =50μm). (C) Representative immunoblots showing effect of MB5 on α1 NKA and Myc expression and Src activation in PC3, *p<0.05 and **p<0.01 as indicated (N=3, Students t test) (D) Representative blot of cytoplasmic and nuclear fractions on EMT markers in PC3 cells with MB5 treatment(N=3). (E) Representative blot showing effect of 1μM MB5 treatment on EMT phenotype in α1 KD cells derived from C4–2, *p<0.05 and **p<0.01 as indicated (N=3, Students t test).; α1 NKA, Na/K-ATPase α1

MB5 reduces tumor growth in the xenograft NOD/SCID mouse model.

The therapeutic potential of MB5 in a tumor xenograft model was first tested by injecting DU145 or corresponding α1 KD cells into the right or left flank of 10-week-old NOD/SCID mice. The tumor growth was monitored twice weekly. Once the tumors reached a volume of approximately 100mm³, the animals received daily injections of MB5 at 20mg/kg intraperitoneally for 3.5 weeks. MB5 treatment significantly reduced (by about 70%) tumor growth in both groups as shown in Fig. 8a, without significantly affecting bodyweight (Fig. s3d). Next, we determined whether MB5 could decrease tumor growth of highly aggressive subclones derived from DU145 in Fig 1. As shown in Fig. 8b, sub-clone 2 formed significantly larger tumors than sub-clone 4 when xenografted into NOD/SCID mice. MB5 treatment (10mg/kg) significantly reduced tumor growth from both cell types (Fig. 8c). Furthermore, analyses of subclone 2 tumor lysates (Fig. 8d) confirmed that MB5 worked by increasing α1 NKA and E-cadherin expression. Fig. s3e summarizes the effect of MB5 treatment on all types of xenografted tumor growth from DU145 and derivative cell lines.

Fig. 8. Effect of MB5 treatment on xenografted tumor growth in NOD/SCID mice.

(A) Effect of MB5 treatment (20mg/kg/day) on tumor growth in NOD/SCOD mice xenografted with DU145 or α1 KD cells. Quantitative analysis of tumor weight (bottom) and representative images of tumors with/without MB5 treatment (top). *p<0.05, ***p<0.001 as indicated (Students’ t test) (N=10 mice per group). (B) Tumor volume and weight of xenografted subclone 4 and 2 cells, ***p<0.001 (one-way ANOVA), *p<0.05 (Students’ t test, N=10). (C) Effect of MB5 treatment (10mg/kg/day) on xenografted tumor growth from aggressive sub-clones 4 and 2. ***p<0.001 and *p<0.05 as indicated (Students t test). N=10 tumors per group. Tumor weight (bottom) *p<0.05 (Students t test). (D) Protein expression analyses of tumor lysates (from sub-clone 2) by Western blot analyses. ***p<0.001 and *p<0.05 as indicated (one-way ANOVA), N=4–6 tumors per group. ANOVA= analysis of variance; KD, knockdown

DISCUSSION

Molecular targets and therapeutics approaches focused on EMT have a high potential in the treatment of metastatic PCa. This study provides the first evidence that a reduction of α1 NKA polypeptide is sufficient to induce EMT, increase invasiveness and consequently aggressiveness of PCa. Upregulation of α1 NKA through gene-overexpression is sufficient to reduce tumor growth in the mouse NOD/SCID model, validating α1 NKA expression as a potential novel target in PCa. Mechanistically, increased endocytosis through activation of α1 NKA/Src signalosome complex is identified as the mechanism underlying the post translational downregulation of NKA in PCa cell lines. High-throughput screening and pharmacological characterization identified the small molecule MB5 as a novel inverse agonist of α1 NKA/Src receptor complex, a blocker of ouabain-induced signal transduction and α1 NKA endocytosis at far lower concentrations than those required to significantly inhibit NKA enzyme function (IC50 for Na/K-ATPase activity =10μM). MB5 effectively reversed EMT and reduced metastatic potential of PCa cells in 3D culture and tumor growth in the mouse tumor xenograft model.

Mechanistically, we suggest that tumor suppressor α1 NKA acts as guardian of the upstream signaling pathways by regulating Src kinase, a protein that is required for receptor tyrosine kinase signaling. In several PCa models, this regulation is attenuated because of increased endocytosis of α1 NKA. We show here that the progressive loss of α1 NKA further aggravates PCa phenotype by promoting EMT through direct inhibition of E-cadherin and occludin expression and dissolution of the cell-cell junction. Evidence from both cell and animal models indicate that the loss of E-cadherin promotes tumor progression, invasion and metastasis [35–37]. We found that α1 NKA is a critical regulator of E-cadherin expression in PCa. This regulation most likely occurred at both transcriptional and posttranslational level. First, α1 NKA downregulation resulted in an increase in Src activity, which could enhance the endocytosis and degradation of E-cadherin [33,38]. This is also consistent with the data presented in Fig. s3a. A second level of regulation may come through transcriptional regulators such as ZEB1 and SNAIL that are known repressors of E-cadherin transcription [39]. Although the exact mechanism is cell-specific, there was a generalized upregulation of mesenchymal markers combined with decrease in adherens junction proteins in all PCa cell lines studied.

Several studies have indicated that NKA can itself function as a cell-cell attachment molecule through NKA β -β interaction between adjacent cells [40]. Specifically, treatment of LLC-PK1 cells with TGFβ was previously shown to induce an EMT phenotype by downregulating NKA β subunit expression through a posttranslational mechanism [12]. However, β subunit itself does not have any known catalytic or Src-mediated signaling function. We suggest that activation of α1 NKA/Src signaling complex in cancer cells contributes to its decreased expression at the plasma membrane. Factors that are common in the tumor microenvironment, such as hypoxia/oxidative stress, can induce endocytosis of the NKA through a NKA/Src-dependent feedforward mechanism known as the NKA amplification loop [41][42–44]. Furthermore, increased extracellular potassium released from apoptotic and necrotic cancer cells [45] may also stabilize the α1 NKA/Src signaling complex in an active state similar to the one stabilized by ouabain, and thereby promote endocytosis. We propose that this can activate multiple oncogenic signaling pathways and also lead to weakened cell-cell attachment by downregulating NKA β-β interaction (Fig. 4).

In this respect, MB5 as an inverse agonist of the receptor α1 NKA/Src, potently blocked the endocytosis and increased the surface expression of α1 NKA in PCa cells. MB5 was also effective in reversing the EMT phenotype by upregulation of E-cadherin and down-regulation of mesenchymal markers SNAIL, SLUG and ZEB1. Consequently, it inhibited the growth and invasiveness of PCa spheroids. Finally, xenograft studies confirmed that MB5 effectively reduced tumor growth of PCa. Two aspects of this new discovery are noted. First, to the best of our knowledge, MB5 represents the first class of inverse agonists of receptor NKA/Src complex. Our findings demonstrate the need and feasibility for developing other potent, effective and structurally diverse classes of inverse agonists targeting the α1 NKA/Src signaling complex. Moreover, MB5 could serve as a prototype to generate potential anti-cancer drug candidates. Second, in addition to PCa, the loss of α1 NKA occurs in several other types of epithelia-derived tumors [14,15,46]. Therefore, it will be important to test whether the mechanism identified here also applies to these cancers, and if MB5, as well as other inverse agonists of the NKA receptor could be as effective as potential therapeutics. Moreover, as increased endocytosis of α1 NKA is a key mechanism for losing dynamic regulation of Src in the plasma membrane, it is equally important to understand whether the cellular distribution of α1 NKA is altered by endocytosis in other types of cancers.

CONCLUSIONS

Our study identified α1 NKA loss as a significant mechanism of EMT and increase in metastatic potential of PCa. This mechanism is based on increased activation and subsequent endocytosis of the α1 NKA/Src receptor complex in cancer cells and can be reversed by inverse agonists of this receptor. Our findings, therefore, lay the groundwork for developing therapeutics against metastatic PCa, by using MB5 as a prototype.

METHODS

Cell lines

DU145, PC3 and C4–2 cell lines were purchased from and maintained according to ATCC recommendations and monitored regularly for Mycoplasma contamination. Parental DU145 and derived cell lines were cultured in high-glucose DMEM medium supplemented with 10% Fetal Bovine Serum and 1% Penicillin/Streptomycin in 37°c humidified incubator with 5% CO₂. LLC-PK1, PY17, Y260A, A425P and YFP-α1TCN cells were cultured in the same media and under similar conditions. PC3 and C4–2 cell lines were cultured in Roswell park memorial institute (RPMI) medium with similar conditions as described above. α1 knockdown (KD) cells were generated from DU145 and C4–2 using a α1 NKA –specific small interference RNA (siRNA), as previously described [15]. Knockdown was verified by both quantitative polymerase chain reaction (qPCR) and Western blot analyses. Rat α1 NKA rescued cell lines were generated by transfecting KD cells with a pRC/CMV-α1 AACm1 vector followed by selection with ouabain (2 μM), as we have previously described[19]. Cells were passaged for three generations without ouabain before conducting experiments.

Mice studies

Animal protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of University of Toledo according to NIH guidelines. Tumor xenografts were established by subcutaneous injection of 5 × 10⁶ cancer cells into the left and right flanks of 10-week-old male NOD/SCID mice (Charles River). Tumor length (L) and width (W) were measured with calipers weekly and tumor volume was estimated as V (L × W²)/2. After the tumor volumes reached approximately 100mm³, mice were sacrificed and tumors were harvested. Half of the tumors were used for protein extraction and the other half were digested with collagenase Ι solution (SIGMA-Aldrich) to isolate cancer cells.

For MB5 treatment, after tumor volume reached approximately 100mm³, mice were injected peritoneally with DMSO or MB5 at 20 or 10mg/kg daily and further monitored for tumor growth weekly. At the end of treatment period (about 2–3 weeks), tumors were harvested and tumor lysates were analyzed for protein expression by Western blot analyses.

Antibodies and Reagents

Antibodies were sourced and used as follows:

Antibody	Property	Supplier	Catalogue number	dilution
α1 NKA antibody (α6F)	Mouse monoclonal	Developmental Studies Hybridoma Bank of University of Iowa (Iowa)	a6f	1:1000
PhosphoSrc (Tyr419) antibody	Rabbit polyclonal	Invitrogen	44–660G	1:1000
c-Src B-12 antibody	Mouse monoclonal	Santacruz Biotechnology	sc-8056	1:1000
Rat α1 NKA antibody	Rabbit polyclonal	Dr. T.A.Pressley (Texas Tech University, TX)	Not applicable	1:1000
Anti-phosphotyrosine antibody, clone 4G10	Mouse monoclonal	EMD Millipore	05–321	1:1000
c-Myc antibody		Santacruz Biotechnology		1:1000
Anti-tubulin antibody	Mouse monoclonal	SIGMA	T5168	1:2000
Cyclin D1 antibody	Rabbit monoclonal	Cell Signaling Technology	2978S	1:1000
Cyclin E1 antibody	Rabbit monoclonal	Cell Signaling Technology	20808S	1:1000
p53 antibody	Mouse pantropic	Calbiochem	OP43	1:1000
p21 antibody	Rabbit polyclonal	Santacruz Biotechnology	sc-397	1:1000
Phospho MAPK antibody	Rabbit	Cell Signaling Technology	9101	1:1000
ERK1/2 antibody	Rabbit polyclonal	Santacruz Biotechnology	sc-94	1:1000
Phospho-FAK (Tyr576/7) antibody	Rabbit polyclonal	Cell Signaling Technology	3281S	1:1000
FAK antibody	Rabbit polyclonal	Cell Signaling Technology	3285S	1:1000
Phospho-FAK (Tyr 397) antibody	Rabbit polyclonal	Cell Signaling Technology	3283S	1:1000
Anti-Na+/K+-ATPase β1 antibody clone 464.8	Mouse monoclonal	EMD Millipore	05–382	1:1000
E-cadherin (24E10) antibody	Rabbit monoclonal	Cell Signaling Technology	3195S	1:1000
Anti β-catenin antibody	Mouse monoclonal	BD Bioscience	610153	1:1000
ZO-1 antibody	Rabbit polyclonal	Thermo-Fisher Scientific	61–7300	1:1000
ZO-2 antibody	Rabbit polyclonal	Thermo-Fisher Scientific	38–9100	1:1000
Occludin antibody (OC-3F10)	Mouse monoclonal	Thermo-Fisher Scientific	33–1500	1:1000
SNAIL (C15D3) antibody	Rabbit monoclonal	Cell Signaling Technology	3879S	1:1000
TCF8/ZEB1 antibody	Rabbit monoclonal	Cell Signaling Technology	3396S	1:1000
Vimentin (D21H3) antibody	Rabbit monoclonal	Cell Signaling Technology	5741S	1:1000
MMP-2 antibody	Rabbit monoclonal	Cell Signaling Technology	87809S	1:1000
MMP-9 antibody	Rabbit monoclonal	Cell Signaling Technology	13667S	1:1000
PCNA antibody	Mouse monoclonal	Santacruz Biotechnology	sc-56	1:2000
Lamin B antibody	Goat polyclonal	Santacruz Biotechnology	sc-6216	1:1000
β actin antibody	Mouse monoclonal	Santacruz Biotechnology	sc-47778	1:1000
SLUG (C19G7) antibody	Rabbit monoclonal	Cell Signaling Technology	9585S	1:1000
N cadherin (D4R1H) antibody	Rabbit monoclonal	Cell Signaling Technology	13116S	1:1000

Rat α1 NKA- specific antibody (NASE) was a kind gift from Dr. T. A. Pressley (Texas Tech University, TX) All reagents were obtained from Sigma-Aldrich except FAK inhibitor (Cat. No. 324877; Millipore). MB5 was chemically synthesized by HD Biosciences Co.

Western blot analysis, immunoprecipitation, and immunostaining

Cells were grown to 100% confluency and Western blot were performed as described before [47]. Images were quantified with ImageJ software from NIH. Immunoprecipitation studies were performed as described before [15]. α1 NKA immunostaining was performed by growing cells on sterilized coverslips in 6 well tissue culture plates and permeabilization/ fixation with ice-cold methanol followed by blocking with 5% horse serum and 0.1% Triton X-100 in 1X Phosphate buffered saline (PBS) for 30 minutes. Coverslips were then stained with an anti-α1 NKA antibody (Cat. No.05–369, Millipire) at 1:100 dilution in 1% Bovine Serum Albumin containing 1X PBS solution for overnight. Next day after three washes, coverslips were stained with Alexa Fluor 488 conjugated anti-mouse secondary antibody (Thermo-Fisher Scientific) for 1 h, washed extensively and then imaged using a fluorescent microscope with GFP filter or confocal microscope (LEICA-DMIRE2). E-cadherin and occludin immunostaining was performed in a similar manner. Phospho-Src, phospho ERK1/2 immunostaining were performed as described before [26] and images were taken with a LEICA DMIRE2 confocal microscope.

RNA extraction, cDNA synthesis and qPCR

Total RNA from cells were extracted using RNeasy minikit from Qiagen. Same amount of RNA was used to synthesize cDNA with Superscript III First-Stand Synthesis SuperMix for qRT-PCR (ThermoFisher). qPCR was performed as described before [15].

Boyden chamber assay

Different sub-clones were grown up to 100% confluency and then gently trypsinized with 0.05% trypsin-EDTA and 100,000 cells were plated in upper chamber of a 0.8μ pore containing transwell filter with 0.5% FBS containing media. Full serum (10% FBS) containing media was added to the lower chamber as chemoattractant. After 16 hours, a colorimetric assay was used to determine the migration of cells to the lower side of the filter, with absorbance read at 590nm. Briefly, lower side of the transwell were washed with 1X PBS and then fixed with ice cold methanol for 10 minutes. After gentle washes with 1X PBS, lower side of the filter were stained with Crystal violet solution (0.5% Crystal Violet in 20% Ethanol), washed and the stain was extracted with methanol. Result was normalized against migration of sub-clone 5 cells. For cell migration in presence of pharmacological compounds, cells at 100% confluency were pretreated for 24 hours with or without MB5, PP2 ( Sigma –Aldrich, Catalogue No P0042) or FAK inhibitor I (Millipore, Catalogue No. 324877) [48] and then assay was performed as described above.

3D culture- spheroid formation assay:

Ten thousand cells/well were plated on top of a solidified 3D matrix composed of 1:1 Collagen ( Cat. No. A1048301, Thermofisher Scientific) and Matrigel (Cat. No.354234, Corning) in six well plates and allowed to form spheroids for a week. Full serum (10% FBS) containing DMEM media was added on top of the matrix and was changed every two days. After 7 days, spheroid formation was recorded using a phase contrast microscope fitted with a camera.

3D culture-spheroid invasion assay:

Cells were trypsinized gently (0.05% Trypsin-EDTA) and diluted at 1000 cells in 20 μl of full serum media for DU145 and C4–2 derived sub-clones or 500 cells in 20ul for PC3 cells and allowed to form spheroids using hanging drop technique for 3 days, as described by Berens et al. 2015 [49]. Generated spheroids were then embedded into 3D matrix (not yet solidified at the time of experiment), as described in the previous section, on top of a 1% solidified agarose coating in 48 well plates. Spheroids were monitored from Day 1 to different time points and images were taken using a phase contrast microscope fitted with camera, at the same settings. Spheroid growth or invasion area was quantified using ImageJ software from NIH. The following formulas were used for quantitation of spheroids:

Spheroid invasion = (Invasion area/Spheroid area) *100

Spheroid growth = (Spheroid area at Day 3 or 7-Spheroid area at 16 h)/Spheroid area at 16 h

For drug treatment, compound was diluted into media, from a stock solution. For MMP secretion, conditioned media for 3 days were collected from top of matrix and assessed for MMP secretion by Western blot using Matrix Remodeling Antibody Sampler kit from Cell Signaling Technology (Cat. No.73959).

α1 NKA endocytosis assay:

Endocytosis assay was performed as described before [16]. Briefly, cells in 6 cm dishes were washed three times with ice-cold PBS with 1-mM EDTA and then exposed twice to EZ-link- Sulfo-NHS-S-S-Biotin (Thermo Fisher Scientific) in biotinylation buffer (10 mM triethanolamine, 150 mM NaCl, pH9.0) for 25 min on ice. Unreacted biotin was quenched with 100 mM glycine containing PBS twice. Prewarmed media was then added to internalization dishes and cell surface proteins were allowed to internalize for 1 h at 37°C incubator. Remaining cell surface proteins were cleaved by treatment with 100 mM TCEP and then lysed with lysis buffer (1% Triton- X-100, 150 mM NaCl, 5 mM EDTA, and 50 mM Tris, pH 7.5). Equal amounts of protein from cell lysates were precipitated with streptavidin–agarose beads overnight, washed three times, then eluted with SDS-containing sample buffer, and subjected to immunoblot analysis.

Spheroid aggregation in ultralow attachment plates

Five thousand cells were plated in a six- well ultralow attachment plate from Corning and allowed to form spheroids spontaneously for a week. Cell aggregates were then immunostained for E-cadherin expression using a monoclonal anti-E-cadherin antibody (Cat. No.sc-71008, Santacruz Biotechnology), followed by Alexa-Fluor 488 anti-mouse secondary antibody (Thermo-Fisher Scientific).

Cell Proliferation assay

Cell proliferation assay was performed by plating 5000 cells per well of 96 well plate and cell proliferation was analyzed by Cell Titer Glo Assay (Promega) according to the manufacturer’s instructions.

ATPase activity assay

ATPase activity assay was performed as previously described [26].

Cell death assays

MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay was performed as described before [16]. Cell Titer glow assay was performed in 96 well plates according to manufacturer’s recommendation (Cat. No. G7570, Promega).

Statistical analysis

Data are shown as mean ± SEM. Student’s t-test was used to compare two individual groups and a one-way analysis of variance (ANOVA) followed by multiple comparison analysis via Dunnett’s or Sidak’s test was used when comparing more than two groups. Graphs were prepared and analyzed using GraphPAD PRISM software. Statistical significance was accepted at p<0.05.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request. The authors confirm that all data are included in Section 3 and Supporting Information.