Survival of detached cancer cells is regulated by movement of intracellular Na⁺,K⁺-ATPase

Summary

Beginning of metastasis, cancer cells detach from the primary tumor and they can survive even under loss of anchorage; however, the detachment-elicited mechanisms have remained unknown. Here, we found that Na⁺,K⁺-ATPase α3-isoform (α3NaK) in human cancer cells is dynamically translocated from intracellular vesicles to the plasma membrane when the attached cells are detached and that this mechanism contributes to the survival of the detached (floating) cancer cells. α3NaK was detected in the plasma membrane of floating cancer cells in peritoneal fluids of patients, while it was in the cytoplasm of the cells in primary tumor tissues. On cancer cell detachment, we also found the focal-adhesion-kinase-dependent Ca²⁺ response that induces the α3NaK translocation via nicotinic acid adenine dinucleotide phosphate pathway. Activation of AMP-activated protein kinase was associated with the translocated α3NaK in the plasma membrane. Collectively, our study identifies a unique mechanism for survival of detached cancer cells, opening up new opportunities for development of cancer medicines.

Graphical abstract

Highlights

• •Na⁺,K⁺-ATPase α3-isoform (α3NaK) is localized in cytoplasm of attached cancer cells
• •Intracellular α3NaK is moved to plasma membrane (PM) upon the cell detachment
• •FAK and NAADP-dependent Ca²⁺ response is involved in the translocation of α3NaK
• •Activation of AMPK associated with the PM-α3NaK contributes to the cell survival

Cell Biology; Cancer

Introduction

Cancer metastasis is the major cause of mortality in patients with cancer and responsible for up to 90% of cancer death (Seyfried and Huysentruyt, 2013). During the metastatic process, the detached cancer cells from the primary tumor tissue spread to different sites through blood/lymphatic vessels, then settle, and grow at a site other than the primary site. Malignant floating cancer cells can evade the detachment-induced cell death (DICD) and survive even under the loss of anchorage condition (Buchheit et al., 2012; Guadamillas et al., 2011), whereas normal epithelial cells undergo cell death when they lack their anchorage to the extracellular matrix (ECM). Thus, the anchorage-independent survival is a unique feature of metastatic cancer cells. So far, the AMP-activated protein kinase (AMPK) has been reported to have an essential role in resistance to DICD by activation (phosphorylation) of AMPK (Sundararaman et al., 2016; Jin et al., 2018). However, cell detachment-elicited mechanisms evading DICD have not been fully established.

Na⁺,K⁺-ATPase is mainly located in the plasma membrane (PM) and is a crucial enzyme that regulates membrane potential and cellular ion homeostasis in almost all mammalian cells. On the other hand, abnormal expression of Na⁺,K⁺-ATPase has been reported in various cancers (Durlacher et al., 2015). A number of in vitro and in vivo studies have shown that cardiac glycosides, inhibitors of Na⁺,K⁺-ATPase, can block cancer cell growth (Calderón-Montaño et al., 2014; Diederich et al., 2017; Fujii et al., 2018). In addition, chemical screening showed that some cardiac glycosides exhibited a potent inhibitory effect on the resistance to anoikis (or cell-detachment-induced apoptosis) of cancer cells (Liu et al., 2008). Therefore, Na⁺,K⁺-ATPase is thought to be a therapeutic target for cancer treatment.

Na⁺,K⁺-ATPase α3-isoform (α3NaK) is highly expressed in the PM of neuronal cells and involved in restoring neuronal membrane potential after depolarization and for maintaining neuronal excitability (Holm and Lykke-Hartmann, 2016). On the other hand, α3NaK has been reported to be abnormally expressed in the cells of human colorectal and liver cancers (Sakai et al., 2004; Shibuya et al., 2010). In addition, the expression of α3NaK in intracellular compartments, but not in the PM, was reported in colon and pancreas cancer cell lines (Yang et al., 2009, 2014). However, the pathophysiological roles of the intracellular α3NaK in cancer cells are poorly understood.

In the present study, we found that α3NaK is localized in the intracellular vesicles in attached cancer cells and that the vesicles are translocated to the PM by the loss of cell-ECM adhesion. Interestingly, the translocation of α3NaK contributed to the survival of detached cancer cells.

Results

Loss of cell adhesion induces dynamic trafficking of intracellular α3NaK to the PM

The expression of α3NaK in human cancer tissues was examined by using a human tissue microarray with an anti-α3NaK antibody. The significant expression of α3NaK was observed in a variety of cancers, and percentage of the α3NaK-positive tissues was especially high in colorectal (94%) and gastric (83%) cancers (Figures 1A and S1). In immunohistochemistry of the colorectal cancer and adjacent noncancer tissues from a patient, α3NaK was predominantly expressed in the cytoplasm of the cancer cells, but no significant expression was observed in the noncancer cells (Figure 1B).

Figure 1.

Translocation of α3NaK to the PM by cell detachment

(A) Expression rates (left; red bars) and average expression scores (right; blue bars) of α3NaK were assessed in tissue microarrays of multiple human cancers as shown in Figure S1.

(B) Immunohistochemistry of α3NaK in human colorectal cancer tissue (patient No. 1, red frame) and adjacent noncancer mucosa (patient No. 1, green frame). Scale bars, 500 μm (middle panel) and 50 μm (left and right panels).

(C and D) Localization of α3NaK in the cancer tissue (C) and floating cancer cells in the peritoneal fluid (D) from a patient with gastric cancer (patient No. 2). Cancer cells were identified by staining of CEA. Scale bars, 10 μm.

(E) Localization of α3NaK in floating cancer cells in the peritoneal fluid of a patient with colon cancer (patient No. 3). Cancer cells were identified by staining of CEA. Scale bars, 10 μm.

(F) Fluorescent images of α3NaK (green) and flotillin-2 (red) in attached HT-29 cells. Scale bar, 10 μm.

(G) An immunoelectron image of α3NaK in attached HT-29 cells. Red circles indicate clusters of α3NaK molecules underlying the PM. Blue arrow indicates a microvillus. Scale bar, 100 nm.

(H) Immunocytochemistry using antibodies for α3NaK (green) and flotillin-2 (red) was performed in detached and re-attached HT-29 cells. Cells were detached by the treatment with the solution containing 0.25% trypsin and 10 mM EDTA. Scale bars, 10 μm.

(I) Surface biotinylation assay in HT-29 cells. Western blots of α3NaK, α1NaK, ERp57, and myosin IIA in the cell surface biotinylation samples (surface) and total cell lysates (input) of attached cells (A), detached cells (by 10 mM EDTA [E] or EDTA plus 0.25% trypsin [T]), and reattached cells.

(J) Ouabain (10 μM)-sensitive ⁸⁶Rb⁺ uptake activities of attached and detached HT-29 cells (n = 7). Cells were detached by treatment with the solution containing 10 mM EDTA. ∗∗, p < 0.01.

(K and L) Immunocytochemistry using antibodies for α3NaK (green) and flotillin-2 (red) was performed in attached round-shaped cells which were attenuated their cell-matrix adhesion by treatment with 10 mM EDTA (K) and in detached doublet cells which keep their cell-cell adhesion (L). Scale bars, 10 μm.

During the process of peritoneal dissemination, advanced gastric and colorectal cancer cells detach from their primary locations and thereafter survive even upon anchorage loss. Resistance to the DICD is a crucial factor for the development of peritoneal dissemination of human gastric cancer cells (Yawata et al., 1998). Here, we examined the localization of α3NaK in floating cancer cells in the peritoneal fluid of patients with gastric and colon cancer. Interestingly, localization of α3NaK in the PM was observed in the floating gastric cancer cells which express carcinoembryonic antigen (CEA), a tumor marker of gastrointestinal malignancies (Figure 1D). In contrast, α3NaK was observed only in the cytoplasm of cancer cells in the primary site from the same patient (Figure 1C). In floating colon cancer cells in the peritoneal fluid of another patient, α3NaK was also found to be localized in the PM (Figure 1E). These results prompted us to hypothesize that α3NaK in the cancer cells is translocated from the cytoplasm to the PM by losing the cell anchorage during metastasis.

To explore this hypothesis, we first performed in vitro experiments using human cancer cell lines: colorectal cancer HT-29 cells, gastric cancer MKN45 cells, and hepatocellular carcinoma HepG2 cells. Under the cell-attached condition, α3NaK was detected in the cytoplasm and its localization was not overlapped with flotillin-2, a marker for the PM (Figures 1F and S2A). In addition, the localization of α3NaK was different from organelle markers of the endoplasmic reticulum (ER; calnexin), Golgi body (TGN46), endosome (EEA1), lysosome (LAMP1), and mitochondria (MitoTracker) (Figure S3). Interestingly, most of the α3NaK were colocalized with Rab10, a small GTPase (Stenmark, 2009), which contributes to vesicular trafficking in the cytoplasm of the cancer cells (Figures S4A and S4C). In contrast, no significant colocalization of α3NaK with Rab4, 5, 7, 8, 9, or 11 was observed (Figures S4A and S4B). Electron microscopy analysis with a high-pressure freezing method showed the assembly (less than 50 nm of diameter) of α3NaK underlying the PM of the cells (Figure 1G). The average number of α3NaK molecule in each assembly was around 4, and the average distance of the assembly from the PM was 418 nm (Figure S5A). These results suggest that α3NaK is localized in intracellular Rab10-expressing vesicles in attached cancer cells.

We next examined the changes in the localization of α3NaK upon cancer cell detachment. When cancer cells were detached from the matrix of dishes by treatment with ethylenediaminetetraacetic acid (EDTA) and trypsin, the signal of α3NaK was detected in the PM in which flotillin-2 was localized (Figures 1H, S2B, and S5B), suggesting that α3NaK is translocated from the cytoplasm to the PM upon cell detachment. Interestingly, when the detached cells were replated on a dish, localization of α3NaK in the cytoplasm was found again (Figures 1H, S2B, and S5B). Similarly, localization of Rab10 was changed from the cytoplasm to the PM upon cell detachment, and it was in the cytoplasm again after cell reattachment (Figures S4C and S4D). In addition, we performed surface biotinylation assay. In the assay, no significant bands of intracellular proteins such as the thiol oxidoreductase ERp57 and nonmuscle myosin IIA were observed in the biotinylation samples (Figure 1I), suggesting no leakage of biotin into the cytoplasm through the procedure of cell detachment. Consistent with the results in immunocytochemistry (Figures 1H and S2B), the expression level of α3NaK in the PM (biotinylation samples) of detached cells was much higher than attached and reattached cells (Figures 1I and S2C). On the other hand, the expression level of Na⁺,K⁺-ATPase α1-isoform (α1NaK) in the PM of detached cells was comparable with attached and reattached cells (Figure 1I). In addition, the K⁺-transporting activity of Na⁺,K⁺-ATPase in the PM of the cells was measured using ⁸⁶Rb⁺ as a K⁺ congener. Cardiac glycoside (ouabain)-sensitive ⁸⁶Rb⁺ uptake activity in detached cells was significantly greater than in attached cells, suggesting that α3NaK transferred from the cytoplasm to the PM is functional (Figures 1J and S11A). Furthermore, the localization of α3NaK in the PM was also observed in single cancer cells isolated from human colorectal cancer tissues by enzyme digestion (Figure S6). These results indicate that reversible translocation of α3NaK between the PM and the cytoplasm is mediated by the cell detachment/attachment transition during the metastatic process of cancer cells.

Next, we examined whether the loss of cell-ECM attachment triggers the translocation of α3NaK from the cytoplasm to the PM (PM translocation of α3NaK). Interestingly, the PM translocation of α3NaK was not observed in round-shape attached cells that were partly but not completely have lost their cell-ECM adhesion (obtained by short-time treatment of EDTA; Figures 1K and S5C). In contrast, the PM translocation of α3NaK occurred in the detached cell couplet in which cell-cell adhesion is still retained (Figures 1L and S5C). These results suggest that the loss of cell-ECM but not cell-cell adhesion is required for the α3NaK translocation and that the change in cell curvature during the detachment process is not a trigger of the translocation.

Rab10, nicotinic acid adenine dinucleotide phosphate, and focal adhesion kinase are associated with the PM translocation of α3NaK

Because α3NaK was colocalized with Rab10 in both attached and detached cancer cells (Figure S4), we assessed the effect of Rab10 knockdown on the PM translocation of α3NaK in HT-29 cells. The total expression level of Rab10 was dramatically reduced in the Rab10-knockdown cells (Figure 2A). Interestingly, the surface expression level of α3NaK was significantly decreased in the detached Rab10-knockdown cells, while the total expression level of α3NaK was not changed (Figures 2B and 2C). Rab-GTPases-containing vesicle transport is mediated by actin filaments and microtubules (Hammer third and Wu, 2002). We then examined the effect of latrunculin B, an inhibitor of actin filament polymerization, on the detachment-induced PM translocation of α3NaK. As expected, latrunculin B significantly suppressed the expression level of α3NaK but not α1NaK in the PM of detached cells (Figures 2D, 2H, and S7). In addition, the PM translocation of α3NaK but not α1NaK was blocked when the cells were detached in cooled bathing solution (∼4°C) (Figures 2E, 2H, and S7). These results suggest that the Rab10-related vesicular trafficking is involved in the PM translocation of α3NaK.

Figure 2.

Detachment-induced α3NaK translocation is mediated by Rab10, NAADP, and FAK

(A and B) Cell surface biotinylation in detached HT-29 cells transfected with Rab10 siRNA (si-Rab10) or negative control siRNA (si-NC). Western blots (WB) of Rab10 and α3NaK in the total lysates (input) and biotinylation samples (surface). β-actin was used as a loading control. n = 4. ∗∗, p < 0.01.

(C) Immunocytochemistry using antibodies for α3NaK (green) and flotillin-2 (red) was performed in detached Rab10-knockdown HT-29 cells. Scale bar, 10 μm.

(D and E) Effect of latrunculin B (10 μM, 1h), oleandrin (OLD; 1 μM, 1 h), ouabain (OAB; 1 μM, 1 h), or cooled bath solution (~4°C) on the surface expression level of α3NaK in detached HT-29 cells.

(F) Effect of ouabain (1 μM, 2 min) on the surface expression level of α3NaK in detached cells.

(G) Effects of BAPTA-AM (100 μM, 30 min), thapsigargin (10 μM, 1 h), CPA (30 μM, 1 h), U-73122 (5 μM, 10 min), 2-APB (100 μM, 1 hr), xestospongin C (4 μM, 1 h), Ned-19 (100 μM, 30 min), bafilomycin A1 (100 nM, 1 h), and PF573228 (500 nM, 1 h) on the surface expression level of α3NaK in detached HT-29 cells.

(H) Quantification of the surface expression level of α3NaK in D-G. n = 4–5. ∗∗, p < 0.01. lat. B: latrunculin B; TG: thapsigargin; Xest: xestospongin C; baf: bafilomycin A1; PF: PF573228.

The Na⁺,K⁺-ATPase β-subunit has essential roles in the trafficking and function of the α-subunit. To identify the β-subunit isoform for α3NaK in human cancer cells, α3NaK was immunoprecipitated using the anti-α3NaK antibody in the lysate of HT-29 cells. Shotgun mass spectrometry with the elution fraction detected Na⁺,K⁺-ATPase β1-isoform (β1NaK) with high protein scores (Figure S8). In contrast, no scores of β2 and β3 isoforms were detected in the immunoprecipitation samples. Thus, α3NaK may be coupled with β1NaK in the cells. Similarly, it has been reported that metastasized colorectal cancer cells express the α3NaK-β1NaK complex and that the complex may potentially serve as a novel exploratory biomarker of colorectal cancer metastatic cells in the liver (Baker Bechmann et al., 2016).

We next examined the effects of Na⁺,K⁺-ATPase inhibitors (cardiac glycosides; ouabain and oleandrin) on the detachment-induced PM translocation of α3NaK. Treatment of ouabain and oleandrin at 1 μM for 1 h significantly decreased the expression level of α3NaK (but not α1NaK) in the PM of the detached cells (Figures 2E, 2H, and S7). In contrast, short-time treatment of ouabain (at 1 μM for 2 min) showed no significant change of the expression level of α3NaK in the PM of the detached cells (Figures 2F and 2H). In the experiment to test the permeability of ouabain across the PM, uptake of [³H]-ouabain into HT-29 cells was measured (Figure S9). The [³H]-ouabain uptake was increased in a time-dependent and temperature-dependent manner and reached a maximum at 10 min (Figure S9). However, [³H]-ouabain uptake was very low in 2-min treatment (Figure S9). These results suggest that function (activity) of intracellular α3NaK may be important for the PM translocation of α3NaK.

Intracellular Ca²⁺ acts as a pivotal regulator of various cellular functions including vesicle trafficking and mechanotransduction (Brunger, 2001; Hay, 2007; Iqbal and Zaidi, 2005). Interestingly, we found that the detachment-induced PM translocation of α3NaK was inhibited by Ca²⁺-chelating agent BAPTA-AM and sarco/ER Ca²⁺-ATPase (SERCA) inhibitors, thapsigargin, and cyclopiazonic acid (CPA) (Figures 2G and 2H). These compounds had no effects on the expression level of α1NaK in the PM of the cells (Figure S7). Inositol 1,4,5-triphosphate (IP₃) and phospholipase C (PLC) are related to the Ca²⁺ release from ER through IP₃ receptor. However, the PM translocation of α3NaK was not blocked by IP₃ receptor inhibitors, 2-aminoethoxydiphenyl borate (2-APB) and xestospongin, and a PLC inhibitor U73122 (Figures 2G and 2H). On the other hand, nicotinic acid adenine dinucleotide phosphate (NAADP) is one of the second messengers that release Ca²⁺ from acidic stores such as the endosome and lysosome (Galione et al., 2011). Interestingly, NAADP antagonist Ned-19 (Naylor et al., 2009) that blocks NAADP-mediated Ca²⁺ signaling significantly reduced the PM translocation of α3NaK in the cells (Figures 2G and 2H). Bafilomycin A1, an inhibitor of V-ATPase in the endosome and lysosome, had no effect on the translocation (Figures 2G and 2H). These results suggest that intracellular Ca²⁺ stores stimulated by NAADP but not IP₃ is involved in the PM translocation of α3NaK translocation, whereas the endosome and lysosome are unlikely related to the mechanism.

Focal adhesion is a best-characterized cell-ECM adhesion structure (Worth and Parsons, 2008). Focal adhesion kinase (FAK) is a cytoplasmic tyrosine kinase localized at focal adhesions and plays a key function for integrin-mediated signal transductions (Sulzmaier et al., 2014). In the suspension culture of cancer cells, activation of FAK is related to resistance to DICD (Kočí et al., 2011; Liu et al., 2008). Here, we found that PF573228, an inhibitor of FAK, significantly attenuated the expression level of α3NaK in the PM of the detached HT-29 cells (Figures 2G and 2H). In contrast, it had no effect on the expression level of α1NaK in the PM (Figure S7).

Cell detachment induces a unique Ca²⁺ response

To examine whether the detachment stimulus induces intracellular Ca²⁺ mobilization in cancer cells, the intracellular Ca²⁺ level ([Ca²⁺]_i) in HT-29 cells was measured with a Ca²⁺-sensitive dye Fluo-4. In the experiment under extracellular Ca²⁺-free solution, the weakly attached cell on the dish was clamped (touch stimulation) and subsequently lifted (detach stimulation) using a glass pipette. As previously reported (Hansen et al., 1995; Moerenhout et al., 2001), transient increase in [Ca²⁺]_i was observed by the touch stimulation (Figures 3A and 3B). Interestingly, subsequent detach stimulation increased [Ca²⁺]_i again (Figures 3A and 3B). Given that these increases in [Ca²⁺]_i were observed in Ca²⁺-free solution, the elevations after the touch or detach stimulation seemed to be due to Ca²⁺ release from intracellular Ca²⁺ stores. Indeed, these [Ca²⁺]_i elevations were blocked by thapsigargin and CPA (Figures 3A and 3B). The [Ca²⁺]_i elevation by touch stimulation was significantly attenuated by 2-APB and U73122, whereas the detachment-induced [Ca²⁺]_i elevation was not (Figures 3A and 3B). In contrast, Ned-19 significantly inhibited only the detachment-induced [Ca²⁺]_i increase (Figures 3A and 3B). Bafilomycin A1 affected the [Ca²⁺]_i elevation by neither touch nor detach stimulation (Figures 3A and 3B). Notably, the region where [Ca²⁺]_i was elevated by detach stimulation was distinct from that by touch stimulation (Figure 3C). These results suggest that the touch stimulation triggers IP₃-dependent Ca²⁺ release from the ER, whereas detachment-induced Ca²⁺ release mediated by NAADP is originated from the Ca²⁺ store other than the ER, endosome, and lysosome. In addition, oleandrin significantly attenuated the [Ca²⁺]_i increase by detach stimulation but not touch stimulation (Figures 3A and 3B), suggesting that function (activity) of α3NaK may be involved in the Ca²⁺ response.

Figure 3.

Unique [Ca²⁺]i increase by detachment of cancer cells

(A) [Ca²⁺]_i imaging in HT-29 cells using Fluo-4. A cell was clamped (touch-stimulation) using a glass pipette in the Ca²⁺-free solution. After the increase in [Ca²⁺]_i returned to basal level (~3 min), the cell was lifted (detach-stimulation). Effects of thapsigargin (10 μM, 1 h), CPA (30 μM, 1 h), U73122 (5 μM, 10 min), 2-APB (100 μM, 30 min), Ned-19 (100 μM, 30 min), bafilomycin A1 (100 nM, 1 h), oleandrin (1 μM, 30 min), and PF573228 (500 nM, 30 min) on the fluorescence of Fluo-4 under touch and detach stimulations.

(B) Fluo-4 fluorescence was scored as 0 (not significantly change), 1 (moderate increase), and 2 (marked increase) under touch or detachstimulations. n = 46 (control), n = 11–18 (drug treatments). ∗∗, p < 0.01 versus control.

(C) The region of [Ca²⁺]_i increase in the touch- and detach-stimulations was visualized.

(D) Effects of ECM solution (10 μM collagen and 10 μM fibronectin) on the fluorescence under touch and detach stimulations.

(E) Effects of ECM solution on the cell surface expression of α3NaK and α1NaK in the detached HT-29 cells. n = 3. ∗∗, p < 0.01; NS, p > 0.05.

PF573228 (FAK inhibitor) significantly inhibited the detachment-induced [Ca²⁺]_i increase but not the touch-induced [Ca²⁺]_i increase (Figures 3A and 3B). Interestingly, the addition of ECM compounds (fibronectin and collagen) to the extracellular solution also inhibited the detachment-induced [Ca²⁺]_i increase and the PM translocation of α3NaK but not of α1NaK (Figures 3D and 3E). These results suggest that the Ca²⁺ response is mediated by the activation of FAK induced by loss of cell-ECM interaction.

SERCA3 is involved in the PM-translocation of α3NaK

SERCA pumps play a major role in the Ca²⁺ uptake into the intracellular Ca²⁺ stores. So far, three isoforms, SERCA1-3, have been identified. In the HT-29, HepG2, and MKN45 cells, expressions of SERCA2 and SERCA3 but not SERCA1 were observed (Figure 4A). SERCA3 was mainly colocalized with α3NaK in the cytoplasm of the attached cells (Figure 4B), whereas SERCA2 was not (Figure S10). In the surface biotinylation assay, SERCA3 as well as α3NaK was found to be translocated to the PM by cell detachment, while it was returned to the cytoplasm by the cell reattachment (Figure 4C). Knockdown of SERCA3 using the corresponding siRNA inhibited the detachment-induced [Ca²⁺]_i increase (Figure 4D) and the PM translocation of α3NaK without affecting the total expression (input) level of α3NaK (Figure 4E). On the other hand, both total and surface expression levels of α1NaK were not significantly changed by the transfection of the SERCA3 siRNA (Figure 4E). These results suggest that the cell detachment elicits NAADP-dependent Ca²⁺ release from the SERCA3-expressing Ca²⁺ store.

Figure 4.

Involvement of SERCA3 in the detachment-induced events

(A) Expression of SERCA1, 2, and 3 in HT-29, HepG2, and MKN45 cells. As a positive control for SERCA1, human skeletal muscle was used. β-actin was used as a loading control.

(B) Immunocytochemistry using antibodies for α3NaK and SERCA3 was performed in attached HT-29 cells. Scale bar, 10 μm.

(C) Western blots of SERCA3, α3NaK, and α1NaK using the biotinylation samples (surface) and total lysates (input) of attached (A), detached (D), and re-attached (R) HT-29 cells. n = 5.

(D) Effect of the SERCA3 knockdown (si-SERCA3) on the [Ca²⁺]_i increase induced by touch and detach stimulations. As a control, cells transfected with negative control siRNA (si-NC) were examined. A cell was clamped (touch stimulation) using a glass pipette in the Ca²⁺-free solution. After the increase in [Ca²⁺]_i returned to basal level (~3 min), the cell was lifted (detach-stimulation). n = 19. ∗∗, p < 0.01 vs si-NC.

(E) Western blots of SERCA3, α3NaK, and α1NaK using the biotinylation samples (surface) and total lysates (input) of detached HT-29 cells transfected with siRNA for SERCA3 (si-SERCA3) or si-NC. β-actin was used as a loading control. n = 5. ∗∗, p < 0.01; NS, p > 0.05.

FAK- and NAADP-dependent vesicle exocytosis is induced by cancer cell detachment

To examine whether intracellular vesicle is fused to the PM by the cancer cell detachment, we measured the membrane capacitance of HT-29 cells using whole-cell patch-clamp technique (Neher and Marty, 1982). Elevation of membrane capacitance corresponds to an enhancement of exocytosis (Kilic, 2002). In the attached cells, no significant change of membrane capacitance was observed (Figures 5A and 5G). Interestingly, the cell detachment significantly increased the membrane capacitance (Figures 5B and 5G), and the increase was disappeared under cold (∼4°C) conditions (Figures 5C and 5G). In addition, Ned-19, PF573228, and oleandrin also significantly attenuated the detachment-increased membrane capacitance (Figures 5D–5G). These results suggest that FAK-dependent exocytosis of the α3NaK-expressing vesicles may be elicited by the NAADP-dependent Ca²⁺ response upon cell detachment.

Figure 5.

Increase in membrane capacitance upon cancer cell detachment

(A) Representative trace of membrane capacitance of attached HT-29 cells.

(B–F) The change in membrane capacitance by the cell detachment was measured. The control experiment is shown in (B). In (C), the cell was detached in a cooled bath solution. The cells were pre-treated with 100 μM Ned-19 (D), 500 nM PF573228 (E), and 1 μM oleandrin (F). Red arrows indicate the time point of the cell detachment. Representative traces were shown.

(G) The average rates of changes in membrane capacitance were shown. n = 6–11. ∗∗, p < 0.01.

α3NaK contributes to survival of the detached cancer cells

Metastatic cancer cells acquire resistance to DICD (Buchheit et al., 2012; Seyfried and Huysentruyt, 2013). To investigate whether the PM translocation of α3NaK is required for the survival of the detached cancer cells, α3NaK was knocked down by RNA interference (using α3NaK-siRNA #1) in HT-29 cells. The expression level of α3NaK was significantly decreased in α3NaK-knockdown cells (Figure 6A). In fact, the increase in the ouabain-sensitive ⁸⁶Rb⁺ uptake activity upon detachment (Figure 1J) was inhibited in α3NaK-knockdown cells (Figure S11A).

Figure 6.

α3NaK is involved in survival of the detached cancer cells

(A) Western blots in HT-29 cells transfected with α3NaK-siRNA (#1) (si-α3NaK), α3NaK-siRNA (#1) (si-α3NaK) plus α3NaK-expression vector (O/E), or negative control siRNA (si-NC). The expression level of α3NaK was normalized to that of β-actin (n = 7). ∗∗, p < 0.01.

(B) Cell viability of attached and detached HT-29 cells. The viability in α3NaK-siRNA (#1)-transfected cells was compared with α3NaK-siRNA (#1) plus α3NaK-expression vector (O/E)-transfected cells and si-NC-transfected cells (n = 7). ∗∗, p < 0.01. NS, p > 0.05.

(C) Expression of α3NaK in the human α3NaK-transfected and the empty vector-transfected (mock) colon 38 cells. β-actin was used as a loading control.

(D) Western blots in cell surface biotinylation samples (surface) and total lysates (input) of attached and detached α3NaK-transfected colon 38 cells.

(E) Immunocytochemistry of α3NaK and Rab10 in attached and detached α3NaK-transfected colon 38 cells. Scale bars, 10 μm.

(F) Effect of exogenous α3NaK expression on cell viability. Attached and detached colon 38 cells were used. The viability in α3NaK-transfected cells was compared with mock-transfected cells (n = 5). ∗∗, p < 0.01. NS, p > 0.05.

(G) Effect of α3NaK knock down on caspase 3/7 activity. Attached and detached HT-29 cells were used. The activity in α3NaK-siRNA-transfected cells (si-α3NaK) was compared with NC-siRNA-transfected cells (si-NC) (n = 5–7). ∗, p < 0.05. NS, p > 0.05.

(H) Effect of exogenous α3NaK expression on caspase 3/7 activity. Attached and detached colon 38 cells were used. The activity in α3NaK-transfected cells was compared with mock-transfected cells (n = 6). ∗∗, p < 0.01. NS, p > 0.05.

Interestingly, the silencing of α3NaK significantly reduced the viability of the detached cells (assessed at 9 h after detachment) but not of the attached cells (Figure 6B). Transfection of another siRNA for α3NaK (α3NaK-siRNA #2) also decreased the expression level of α3NaK and viability in detached cells (Figures S11B–S11D). In the α3NaK-knockdown cells, the detachment-reduced viability was rescued by overexpression of cloned human α3NaK (Figures 6A and 6B). On the other hand, knockdown of α1NaK significantly decreased the viability of both attached cells and detached cells as expected (Figures S11E–S11G).

Next, we examined the involvement of α3NaK in the survival of the detached cancer cells using a heterologous expression system. The cloned human α3NaK was overexpressed in mouse colorectal cancer colon 38 cells in which no significant expression of endogenous α3NaK was observed (Figure 6C). Surface biotinylation assays suggested that the expression level of α3NaK in the PM of detached cells was greater than the attached cells (Figure 6D). In immunocytochemistry, exogenous α3NaK and endogenous Rab10 were found in the cytoplasm of the attached cells, whereas they were in the PM of the detached cells (Figure 6E). In the attached conditions, there is no significant difference in the cell viability between α3NaK-transfected and empty-vector (mock)-transfected cells (Figure 6F). Interestingly, the viability of the α3NaK-transfected cells was significantly greater than the mock-transfected cells in the detached conditions (assessed at 9 h after detachment) (Figure 6F). In the α3NaK-transfected cells, the increased cell viability in a detached condition (Figure 6F) was significantly attenuated by treatment of Ned-19 (Figure S11H), suggesting the NAADP-dependent PM translocation of α3NaK contributes to the cancer cell survival.

We then measured the caspase 3/7 activity, which is activated during apoptosis in cancer cells. Silencing of α3NaK further increased detachment-induced caspase 3/7 activation in HT-29 cells (assessed at 4 h after detachment; Figure 6G). Conversely, exogenous expression of α3NaK significantly inhibited the detachment-induced caspase 3/7 activation in colon 38 cells (Figure 6H). These results suggest that the PM translocation of α3NaK is responsible for the anchorage-independent survival of cancer cells.

To examine in vivo roles of α3NaK in the detached cancer cells, the α3NaK- or mock-transfected colon 38 cells were isolated from the culture dish. It was confirmed that no significant difference was observed between α3NaK- and mock-transfected colon 38 cells at 20 min after detachment (95.3% ± 1.1% and 93.5% ± 2.1%, respectively), and the cells were subcutaneously injected into mice within 15 min. Subcutaneous tumors were isolated from the mice and measured their wet weights at 5 and 10 days after injection. The weight of the tumor obtained from α3NaK-overexpressing cells was significantly greater than that from mock-transfected cells (Figure 7A). The mRNA of exogenous human α3NaK was detected in the isolated tumor tissues (Figure S12).

Figure 7.

Expression of α3NaK promotes tumor growth and metastasis in vivo

(A) α3NaK- and mock-transfected colon 38 cells were subcutaneously implanted into the flank of mice. Five and 10 days after injection, cancer tissues were dissected, photographed, and weighed. Three independent experiments were conducted with six mice for each group. ∗∗, p < 0.01.

(B) Relative mRNA expression levels derived from the pcDNA4 vector in the α3NaK- and mock-transfected colon 38 cells were assessed by quantitative real-time PCR analysis. (n = 4). NS, p > 0.05.

(C) α3NaK- and mock-transfected colon 38 cells were injected into the tail vein of mice. Seven days after injection, lung was removed and total RNA was prepared. Relative mRNA expression levels derived from the pcDNA4 vector (corresponding to the number of the colon 38 cells) in the lungs were assessed by quantitative real-time PCR analysis. (n = 8). ∗∗, p < 0.01.

(D) The proposed model of the detachment-induced events in cancer cells.

To identify the association of α3NaK with metastasis in vivo, α3NaK- and mock-transfected colon 38 cells were injected into the tail vein of mice. In a control experiment, mRNA expression derived from the transfected vector (pcDNA4) in the colon 38 cells was measured by the quantitative real-time polymerase chain reaction (PCR) analysis. No significant difference in transfection efficiency was observed between the α3NaK- and mock-transfected cells (Figure 7B). Seven days after injection, the lung was removed and total RNA was prepared. Metastasis of cancer cells in the lung was evaluated by quantitative real-time PCR analysis. The amount of the vector (corresponding to the number of the colon 38 cells) in the lungs of mice injected with α3NaK-transfected cells was significantly higher than those injected with mock-transfected cells (Figure 7C). These results suggest that α3NaK contributes to the survival of the detached cancer cells in vivo.

Activation of AMPK is involved in downstream of the PM translocation of α3NaK

How does α3NaK confer survival signaling in the detached cancer cells? Here, we focused on AMPK and reactive oxygen species (ROS). AMPK has been known to have an essential role in anoikis resistance. AMPK is phosphorylated at position 172 (threonine residue) and activated upon matrix deprivation (Sundararaman et al., 2016; Jin et al., 2018). We then examined the association between α3NaK and AMPK in detached cancer cells (Figure 8A). As previous reports, cell detachment activated the phosphorylation of AMPK at threonine 172 in HT-29 cells. Interestingly, knockdown of α3NaK suppressed the detachment-induced phosphorylation of AMPK (Figure 8A). These results suggest that activation of AMPK is associated with the translocated α3NaK for survival of detached cancer cells. On the other hand, increase in intracellular Ca²⁺ and ROS has been reported to be involved in the AMPK phosphorylation (Sundararaman et al., 2016). However, no significant difference in the ROS level was observed in the α3NaK-knockdown cells upon cell detachment (Figure 8B).

Figure 8.

Downstream of the PM-translocation of α3NaK

(A) Effect of α3NaK-knockdown on the AMPK phosphorylation. Expression of pAMPK (Thr172), total AMPK, and β-actin in attached and detached HT-29 cells transfected with siRNA for α3NaK (#1) (α3NaK) or NC-siRNA (NC). The expression level of pAMPK (Thr172) was normalized to that of total AMPK (n = 3). ^∗p < 0.05; ^∗∗p < 0.01; NS, p > 0.05.

(B) Effect of α3NaK-knockdown on the intracellular ROS generation. Intracellular ROS level in HT-29 cells transfected with siRNA for α3NaK (#1) (α3NaK) or NC-siRNA (NC). Cells were treated with EDTA for 5 min and were detached by gently pipetting. The fluorescence was sequentially measured from 0 min (attached cells; before pipetting) to 60 min (n = 4).

Discussion

In this study, we found that α3NaK in intracellular vesicles are dynamically translocated to the PM by loss of anchorage in the cancer cells and that this mechanism is mediated by FAK- and NAADP-dependent Ca²⁺ mobilization (Figure 7D). Patch-clamp capacitance measurements also demonstrated the induction of the vesicle exocytosis by the FAK- and NAADP-dependent pathway upon cancer cell detachment. The expression of α3NaK in the PM was detected in floating cancer cells obtained from peritoneal fluids of patients. Our in vitro and in vivo studies showed that the PM-translocation of α3NaK contributes to the survival of the detached cancer cells.

α3NaK has been thought to be predominantly expressed in the PM of the neuronal cells. However, our tissue microarray analysis showed that the expression of α3NaK was widely observed in the cytoplasm of various types of human cancer cells. In the cells, α3NaK was colocalized with Rab10 which is a key regulator of intracellular vesicle trafficking (Stenmark, 2009). Here, the knockdown of Rab10 significantly inhibited the detachment-induced translocation of α3NaK in the cancer cells. In addition, we found that α3NaK is colocalized with SERCA3 in the cells. SERCA Ca²⁺ pumps are generally involved in maintaining and replenishing internal Ca²⁺ stores. The knockdown of SERCA3 inhibited the detachment-induced Ca²⁺ mobilization and the PM translocation of α3NaK. Cardiac glycoside (oleandrin) inhibited the detachment-induced Ca²⁺ mobilization and subsequent vesicle exocytosis. These results suggest that the vesicles which express α3NaK, Rab10, and SERCA3 can function as a Ca²⁺ store for the detachment-induced Ca²⁺ release (Figure 6H). Interestingly, the Ca²⁺ release was suppressed by inhibitors of NAADP and FAK but not of IP₃R, PLC, and V-ATPase. To our knowledge, such Ca²⁺ store with a unique pharmacological property has not been reported to date. Rab10 was highly expressed in human liver cancer tissues (He et al., 2002; Wang et al., 2017), and the Rab10 overexpression in the cancer tissues correlated with poor prognosis including distant metastasis (Wang et al., 2017). Our results suggest that Rab10 may regulate translocation of the α3NaK and SERCA3-expressing vesicles, resulting in cancer cell malignancy. Furthermore, our proteome analysis indicated α3NaK is coupled with β1NaK (not β2NaK and β3NaK) in the intracellular Rab10-expressing vesicles of HT-29 cells. β-subunit plays a crucial role for the trafficking and functional expression of α-subunit of Na⁺,K⁺-ATPase including α3NaK (Dobretsov and Stimers, 2005). Interestingly, α3NaK is also associated with β1NaK in PM of neuronal cells (Shrivastava et al., 2015). Future studies are needed to clarify why the α3NaK-β1NaK complex is localized in the intracellular vesicles rather than PM of the attached cancer cells. In addition, detachment-induced vesicle exocytosis and PM translocation of α3NaK were significantly inhibited by oleandrin, suggesting that function of the α3NaK-β1NaK complex may play an important role in the translocation.

It has recently been reported that the administration of cardiac glycosides (ouabain and digoxin) remarkably reduced the total metastatic burden in in vivo mouse models (Gkountela et al., 2019): Inhibition of Na⁺,K⁺-ATPase leads to increase in [Ca²⁺]_i, resulting in dissociation of circulating tumor cells (CTCs) clusters which is associated with the increased metastatic potential. In addition, ouabain targets α3NaK to inhibit cell proliferation and induce apoptosis in human OS-RC-2 renal cancer cells and human NCI-H446 small-cell lung cancer cells (Xiao et al., 2017). Bufalin, a cardiac glycoside, induced apoptosis of human T24 bladder carcinoma cells through inactivation of α3NaK (Huang and Zhang, 2018). In the present study, the knockdown and overexpression of α3NaK demonstrate that the PM translocation of α3NaK is involved in the survival of the detached cancer cells, whereas α1NaK unlikely contributes to this mechanism. The ouabain-sensitive ⁸⁶Rb⁺-uptake activity of cancer cells was increased upon cancer cell detachment, suggesting functional expression of α3NaK in the PM. The contribution of α3NaK function in the PM to the formation of the CTCs clustering remains to be elucidated in a future study.

Disassembly of focal adhesions is necessary to cellular migration and FAK is associated with this process (McLean et al., 2005). FAK is normally activated when the cells attach to appropriate matrix proteins. However, it is noted that FAK is activated in human epidermal cancer cells under the detached culture conditions and that the FAK activation could stimulate tumor cell migration and therefore induce metastasis (Katayama et al., 2008). In addition, the administration of an NAADP antagonist Ned-19 strongly reduced the number of lung metastases of B16 melanoma cells in in vivo studies and FAK inactivation is involved in the mechanism (Favia et al., 2016). Thus, the FAK-NAADP axis is thought to be involved in metastasis mechanisms. Here, we found that both Ned-19 and FAK inhibitors significantly block the detachment-induced [Ca²⁺]_i increase and the PM translocation of α3NaK, suggesting the involvement of FAK and NAADP in a unique Ca²⁺ response induced by cancer cell detachment. In fact, Ca²⁺ signals are elicited by external forces such as stretching tension, scratch, compression, shear force, and osmotic pressure. FAK is one of the molecules related to cell-ECM contacts and functions as a mechanosensor translating the force into biochemical signals (Hytönen and Wehrle-Haller, 2016). Therefore, our study may be the first report describing mechanosensing machinery induced by loss of cell-ECM anchorage.

We also found that phosphorylation of AMPK is stimulated in the detached cancer cells and that silencing of α3NaK suppresses the detachment-induced phosphorylation of AMPK. In contrast, no significant difference was observed in intracellular ROS levels in the α3NaK-knockdown cells. These results suggest that activation of AMPK is involved in downstream of the PM translocation of α3NaK. Mechanism of functional relationship between α3NaK and AMPK in the anoikis resistance is an important topic for future research.

In conclusion, we revealed a novel mechanotransduction upon cell-ECM detachment which gives rise to a dynamic translocation of α3NaK to the PM for metastatic cancer cell survival. Therefore, α3NaK may be a potent therapeutic target for inhibiting metastasis of cancer cells.

Limitations of the study

Our in vivo and in vitro analysis demonstrated that NAADP-dependent Ca²⁺-signaling-induced translocation of α3NaK to the PM is involved in cell survival of detached cancer cells. We also showed that the PM translocation of α3NaK was observed in the floating cancer cells in the peritoneal fluid of patients with gastric and colon cancer and the single cancer cells isolated from human colorectal cancer tissues by enzyme digestion. However, the intracellular phenomenon at the moment when cancer cells spontaneously detach from the primary tissue has not been verified for technical and ethical reasons. Furthermore, to expand our findings, it is important to investigate the localization and function of α3NaK in CTCs isolated from patients with cancer.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Hideki Sakai (sakaih@pha.u-toyama.ac.jp).

Materials availability

This work did not generate new unique reagents.

Data and code availability

This article includes all analyzed data.

Methods

All methods can be found in the accompanying transparent methods supplemental file.

Published: May 21, 2021