The development of metastases largely relies on the capacity of cancer cells to invade extracellular matrices (ECM) using two invasion modes termed ‘mesenchymal’ and ‘amoeboid’, with possible transitions between these modes. Here we show that the SCN4B gene, encoding for the β4 protein, initially characterized as an auxiliary subunit of voltage-gated sodium channels (NaV) in excitable tissues, is expressed in normal epithelial cells and that reduced β4 protein levels in breast cancer biopsies correlate with high-grade primary and metastatic tumours. In cancer cells, reducing β4 expression increases RhoA activity, potentiates cell migration and invasiveness, primary tumour growth and metastatic spreading, by promoting the acquisition of an amoeboid–mesenchymal hybrid phenotype. This hyperactivated migration is independent of NaV and is prevented by overexpression of the intracellular C-terminus of β4. Conversely, SCN4B overexpression reduces cancer cell invasiveness and tumour progression, indicating that SCN4B/β4 represents a metastasis-suppressor gene.
The acquisitions of extensive invasion potencies by cancer cells are key components in the metastatic cascade, hence in patients survival1,2. During the last decade, important knowledge in various processes of cancer cell migration and invasiveness has emerged3. In the ‘mesenchymal mode’, engaged cells harbour an elongated fibroblast-like morphology, with a rear-to-front lamellopodial cell polarity, and generate a path in the extracellular matrix (ECM) through proteolytic remodelling. This is performed by invadosomal structures, which are F-actin-rich organelles, protrusive into the ECM and responsible for its proteolysis through the recruitment of both membrane-associated and extracellularly released soluble proteases4,5. In the other invasive mode, called ‘amoeboid’, cancer cells show no obvious polarity but a rounded morphology, and display high potentials for migration and invasiveness6. In this case, strong actomyosin contractions propel the cell, which deforms and squeezes inside small gaps of the ECM, with no need to degrade it. While different cancer cell types may preferentially engage into the mesenchymal mode or the amoeboid one, the most aggressive cancer cells show high plasticity and are able to switch from one phenotype to the other6. Such transitions, orchestrated by RhoGTPases family members7,8,9, offer selective advantages and compensatory mechanisms to migrating cancer cells, presumably abrogating the efficacy of anticancer treatments10. Indeed, attempts to reduce cancer cell invasiveness and metastatic dissemination by targeting proteolytic activity and ECM remodelling have largely failed because of the adaptive compensatory mechanism sustaining protease-independent processes11.
Voltage-gated sodium channels (NaV) are composed of one large pore-forming principal subunit (nine genes encoding nine proteins, NaV1.1–1.9)12,13 and one or two smaller transmembrane subunits considered as auxiliary (four genes SCN1B to SCN4B, generating four subunits, β1 to β4)14. The activity of NaV gives rise to Na+ currents (INa) generating action potentials in excitable cells such as neurons, skeletal and cardiac muscle cells. As a result, these proteins have been considered hallmarks of excitable cells. However, multiple studies have recently demonstrated their expression in non-excitable cells, in which they regulate cellular functions such as migration, differentiation, endosome acidification, phagocytosis and podosome formation15. In addition, NaV are abnormally expressed in carcinoma cells and tumour biopsies, and their activity is associated with aggressive features and cancer progression16,17,18. Expression of the NaV1.5 isoform in breast tumours is correlated with metastases development and patients’ death19,20. In highly aggressive human breast cancer cells, the activity of pore-forming NaV1.5 is not associated with cell excitability but with ECM degradation and cancer cell invasiveness21, hence favouring metastases development22,23. NaV1.5-dependent invasiveness is mediated through allosteric modulation of the Na+–H+ exchanger NHE1, subsequent acidification of the pericellular microenvironment and activation of extracellular acidic cysteine cathepsins24,25,26. Furthermore, NaV1.5 sustains Src kinase activity, polymerization of actin and acquisition by cells of a spindle-shaped elongated morphology26. Altogether, these results indicate a critical role for NaV1.5 in ‘mesenchymal invasion’. The participation of non-pore-forming SCNxB/β subunits in oncogenic processes was not studied as extensively, with the exception of the β1 subunit27, and their roles during metastatic progression remain largely unknown.
In this study, we show that the SCN4B/β4 subunit is expressed in normal epithelial cells and tissues, but is strongly downregulated in aggressive cancer cells and tumours. The loss of SCN4B/β4 enhances cancer cell migration and metastases formation through a RhoA-dependent signalling pathway, independently of pore-forming NaV subunits.
Association with poor prognosis
Expression of the β4 protein, encoded by the SCN4B gene28, has mostly been studied in excitable cells in which mutations have been linked to sodium channelopathies29,30. Initial immunohistochemical analyses performed in normal and cancer breast tissues indicated that the SCN4B/β4 protein was specifically expressed in epithelial cells from normal mammary acini, but was significantly downregulated in cancer cells (Fig. 1a,b). Levels of β4 expression were analysed by immunohistochemistry on tissue microarrays containing normal breast, hyperplasic, dysplasia, breast cancer and lymph node metastases (LNM) samples. Again, the expression of SCN4B/β4 was strong in epithelial cells from normal non-excitable mammary tissues (Fig. 2a,b), as well as in breast hyperplasia (Supplementary Fig. 1). Levels of SCN4B/β4 expression in mammary hyperplasia and dysplasia were similar to those recorded in normal mammary tissues, but were remarkably reduced in biopsies of mammary carcinomas. The most important reductions in SCN4B/β4 expression were observed when comparing in situ grade I to invasive grade II breast tumours. SCN4B/β4 protein was weakly expressed or totally absent in most grade II and III primary tumours studied, as well as in LNM (Fig. 2c,d). Lower levels of the SCN4B gene transcript in breast cancer tissues, compared with non-cancer tissues, were also found to be highly significant in RNA sequence expression analysed from The Cancer Genome Atlas Network (Fig. 2e) and to be associated with an increased risk of metastatic relapse or death in breast cancer patients (Fig. 2f,g). Levels of expression of the other SCNxB genes were also assessed. Expressions of SCN1B, SCN2B and SCN3B were lower in cancer compared with non-cancer tissues (Supplementary Fig. 2a,c,e). However, there was no association between the levels of expression of these genes and the risk of metastatic relapse (Supplementary Fig. 2b,d and f). Importantly, SCN1B and SCN4B genes appeared to be the two most highly expressed SCNxB genes in non-cancer tissues (Supplementary Fig. 2g). Furthermore, SCN4B seemed to be the most significantly downregulated SCNxB gene in breast cancer tissues compared with non-cancer tissues (Supplementary Fig. 2h). Similarly, the analysis of data from two published studies31,32 showed that SCN4B expression levels were downregulated in lung cancer compared with normal lung tissues (Supplementary Fig. 3a,b) and our immunohistochemical analyses in lung cancer tissue microarrays also identified a tendency for decreased protein expression in high-grade primary lung tumours and metastases (Supplementary Fig. 3c,d). SCN4B expression was also down-regulated in prostate, colon and rectal cancers compared with normal tissues (Supplementary Fig. 3e,g). This suggests that SCN4B/β4 is normally expressed in normal epithelial cells (also confirmed by immunohistochemical analyses of normal oesophagus (Supplementary Fig. 4) and that its loss of expression is associated with a gain in invasive properties by cancer cells and an aggressive progression of epithelial tumours. Correlatively, we found that SCN4B expression was higher in non-cancer epithelial mammary MCF-10A compared with several breast cancer cell lines such as MCF-7, MDA-MB-468, MDA-MB-435s and MDA-MB-231 (Fig. 3a,b). Particularly, the expression level of SCN4B was low in the highly invasive and metastatic MDA-MB-231 breast cancer cells, known to express functional NaV1.5 (ref. 22). SCN4B mRNA (Fig. 3c) and protein (Fig. 3d) were expressed in MDA-MB-231 cells genetically modified with the luciferase gene (MDA-MB-231-Luc cells). SCN1B/β1 and SCN2B/β2 were also expressed, but not SCN3B/β3. The lack of expression of SCN3B/β3 was further demonstrated in conventional PCR using two different couples of primers (Supplementary Fig. 5a), thus confirming previously published results obtained with wild-type cells24,33. Transient silencing of SCNxB genes using specific small-interfering RNA (siRNA: siSCN1B, siSCN2B or siSCN4B) was responsible for significant (65–80%) decreases in protein levels 48 h after transfection, as compared with a control null-target siRNA (siCTL) (Fig. 3e). The downregulation of one of the SCNxB gene had no effect on the mRNA expression of the others, suggesting the absence of compensation between SCNxB/β subunits in these conditions (Supplementary Fig. 5b). While reduced expression of either SCN1B/β1 or SCN2B/β2 decreased cancer cell invasiveness through Matrigel-coated invasion chambers by 42.8±6.6% (n=8) and 51.7±0.3% (n=8) (mean±s.e.m. (number of independent experiments)), respectively, inhibition of SCN4B expression enhanced invasiveness by 62.4±12.2% (n=8) (Fig. 3f,g). We also analysed the consequences of knocking-down the expression of SCN4B on MDA-MB-231-Luc invasiveness in vivo using the zebrafish model of micrometastasis34,35. Sixty-one per cent of zebrafish embryos injected with siCTL cells had their organs colonized. This number was increased to approximately 87% of the embryos presenting micrometastases when injected with siSCN4B cells, resulting in an increase in the zebrafish colonization index by 1.41±0.08 fold (n=3) (Fig. 3h,i). These quantifications were performed 72 h after siRNA transfection, a time for which SCN4B downregulation was still efficient (Supplementary Fig. 6a).
Independence of pore-forming NaV subunits
Breast cancer cell invasiveness has been demonstrated to be strongly regulated by the activity of the pore-forming NaV1.5 (refs 19, 20, 24). Therefore, we initially hypothesized that loss of SCN4B expression would increase NaV1.5 activity in highly aggressive cancer cells. To test this hypothesis, we generated MDA-MB-231-Luc-derived cell lines stably expressing a null-target small hairpin RNA (shCTL cells), or expressing shRNAs targeting either the expression of SCN5A gene encoding for NaV1.5 (shSCN5A cells, in which the expression of the SCN4B gene is not changed) as previously described22 or targeting SCN4B transcripts (shSCN4B cells). The use of shSCN4B resulted in 81.1±0.2% (n=37) decrease in mRNA levels (Supplementary Fig. 6b) with no effect on the expression of the other SCNxB subunits (Supplementary Fig. 6d,e). The three cell lines generated displayed identical viability and growth properties (Supplementary Fig. 6c). In shCTL cells, INa is known to be generated by the sole NaV1.5 isoform which, while commonly referred to as tetrodotoxin (TTX)-resistant, can almost completely be inhibited by 30 μM TTX22,36. Similarly to already reported results24,37, the inhibition of NaV1.5 sodium currents in shCTL cells using 30 μM TTX reduced invasiveness by 45.3±4.3% (n=8), similar to the reduction in invasiveness (43.4±5.1%, n=6) in shSCN5A cells, not expressing NaV1.5 and which were no longer sensitive to the addition of TTX (Fig. 4a). Knocking-down the expression of the SCN4B gene with different interfering RNA sequences resulted in similar potentiations of aggressiveness (shSCN4B, Fig. 4a and siSCN4B; Fig. 3e,f). ShSCN4B cell invasiveness was 281.8±16.2% (n=16) higher than that of shCTL cells. The treatment of shSCN4B cells with 30 μM TTX, a concentration that inhibits all NaV channels, with the exception of the very TTX-resistant NaV1.8, significantly reduce their invasiveness (Fig. 4a). To assess a possible independence from NaV1.5 in the increased invasiveness mediated by the loss of SCN4B expression, we silenced SCN5A expression in shSCN4B cells. This significantly reduced cancer cell invasiveness, which was still 184.3±27.0% (n=12) of shCTL cells (Fig. 4b). We then transiently inhibited the expression of SCN4B using siSCN4B siRNA in shSCN5A cells, which no longer express NaV1.5. These cells were also treated or not with (i) 3 μM TTX which inhibits all TTX-sensitive channels (NaV1.1–1.4 and NaV1.6–1.7) or (ii) 30 μM TTX to also inhibit NaV1.5 and NaV1.9 TTX-resistant channels, or (iii) with 30 nM of the specific NaV1.8 inhibitor A80346738. As shown in Fig. 4c, silencing SCN4B expression with siSCN4B in shSCN5A cells increased invasiveness (+251.7±20.8%, n=6) similarly to what was observed with shSCN4B cells expressing NaV1.5 (Fig. 4a). Furthermore, neither TTX nor A803467 reduced invasiveness of shSCN5A cells transfected with siCTL or siSCN4B (Fig. 4c). Altogether, these data clearly demonstrate that the increased invasiveness of SCN4B gene-suppressed cells was not a consequence of the upregulation of NaV1.5 or of any other NaV. We also assessed the effect of knocking-down the expression of SCN4B, using siRNA, on the invasive properties of cancer cell lines, such as the non-small-cell human lung cancer H460 or the human prostate cancer PC3 cell lines, known to express functional NaV (NaV+) contributing to mesenchymal invasion. In addition, we used cancer cell lines that do not express NaV (NaV−) such as the less invasive human breast MDA-MB-468 and the non-small-cell lung A549 cancer cell lines21,39,40. In all cancer cell lines tested, showing similar levels of SCN4B expression (Supplementary Fig. 5d), transfection of siSCN4B significantly increased the invasiveness, from 23.7±8.3% (n=3) increase for PC3 cells up to 60.9±9.5% (n=12) increase in A549 cells (Fig. 4d), confirming that SCN4B/β4 loss potentiates invasiveness independently of the pore-forming NaV subunit.
Because NaV1.5 has been reported as an important regulator of mesenchymal invasion in breast cancer cells through the potentiation of NHE1-dependent H+ efflux and ECM degradation22,25,26, we further investigated its regulation in shSCN4B cells. Patch-clamp recordings revealed, contrary to our initial assumptions, that the loss of SCN4B expression significantly decreased the maximal peak INa, as observed in the INa–voltage relationship (Fig. 5a), suggesting a reduction of channel density at the plasma membrane. This could be due to the reduction of SCN5A expression levels as observed by quantitative PCR (qPCR) (Supplementary Fig. 6d). In shSCN4B cells, the INa activation–voltage relationship was slightly depolarized as compared with shCTL cells (V1/2–activation were −37.2±0.9 mV (n=22) and −40.1±1.0 mV (n=18), respectively, P=0.037, Student’s t-test). More importantly, the availability–voltage relationship was also significantly shifted to a more depolarized potential in shSCN4B cells compared with shCTL cells (V1/2–availability were −80.6±1.6 mV (n=22) and −86.3±1.6 mV (n=18), respectively, P=0.016, Student’s t-test; Fig. 5b), suggesting that even though there were less NaV1.5 channels at the plasma membrane, they might be more active at the membrane potential of cancer cells (comprised between −30 and −40 mV) through an increased persistent window current. Indeed, while the peak INa was reduced in shSCN4B cells, the persistent INa recorded for a membrane depolarization from −100 to −30 mV was strictly identical (Fig. 5c,d). Therefore, the ratio INa persistent/INa peak was significantly enhanced in shSCN4B cells (Fig. 5e). The sensitivity of INa to TTX was unchanged in shSCN4B cells and could almost be fully inhibited by 30 μM TTX (inhibited by 85.6±2.9%, n=7–12; Fig. 5f). Because the persistent INa was similar in shCTL and shSCN4B cells, there was no difference in the efflux of H+ in the two cell lines, measured as previously published25, by the addition of 130 mM NaCl in NH4Cl-pulse-wash-acidified cells in a sodium-free solution (Fig. 5g,h). As a consequence, shSCN4B cells demonstrated identical ECM degradative activities as compared with shCTL cells (Fig. 5i,j), and similar levels of co-localization between the phosphorylated form (Y421) of the actin nucleation-promoting factor cortactin, used as a marker of invadopodial structures41, and DQ-gelatin degradation (Supplementary Fig. 5c). These results suggested that the invadopodial activity was similar in shCTL and shSCN4B cells.
RhoA-dependent amoeboid cell migration
While cells lacking SCN4B expression have the ability to degrade the ECM, the pharmacological inhibition of proteases by GM6001 (MMP inhibitor), leupeptin (cysteine, serine and threonine peptidases inhibitor) or E64 (cysteine cathepsin inhibitor) did not completely prevent the increased invasiveness observed in shSCN4B as compared with shCTL cells (Fig. 6a). The most potent effect was obtained with GM6001, which reduced shSCN4B cell invasiveness by about 32%. This suggested that the enhancement of invasiveness was not solely due to an increase in ECM proteolysis, thus prompting us to analyse the migratory abilities of cell lines using time-lapse cell tracking experiments (Fig. 6b). As expected, the loss of SCN4B expression significantly increased the migration speed (medians were 0.889 μm min−1 for shCTL (n=106) and 1.265 μm min−1 for shSCN4B cells (n=96), P<0.001, Mann–Whitney rank sum test; Fig. 6c), as well as the track length after 3 h measurements (medians were 177.76 μm for shCTL (n=106) and 246.73 μm for shSCN4B cells (n=96), P<0.001, Mann–Whitney rank sum test; Fig. 6d), but with no change in net displacement (Supplementary Fig. 6f). This increase in migration velocity was evocative of a transition towards the amoeboid invasiveness, which is better visualized in three-dimensional (3D) matrices. To study this property more adequately shCTL and shSCN4B cells were seeded into a 3D matrix composed of Matrigel and invasive abilities were analysed, in the absence or presence of GM6001, using time-lapse single-cell tracking over 48 h (Fig. 6e). The loss of SCN4B expression significantly increased the 3D track length travelled by cells over 48 h (medians were 399.39 μm for shCTL and 789.11 μm for shSCN4B, P<0.001, Mann–Whitney rank sum test; Fig. 6f). This increase in 3D invasion could not be attributed to the increase in ECM degradative activity, since shSCN4B invasion was significantly higher than in shCTL cells in the presence of GM6001 (medians were 246.76 μm for shCTL+ GM6001 (n=30) and 379.87 μm for shSCN4B+GM6001 (n=30), P=0.038, Mann–Whitney rank sum test; Fig. 6f). The amoeboid phenotype is generally characterized by a rounded morphology, the presence of blebs at the cell surface and a relative independence of the interaction with the substratum revealed by a decreased number of filopodial structures. Consistent with this, shSCN4B cells demonstrated striking changes in morphology with a higher circularity index (0.48±0.02 (n=88) versus 0.34±0.02 (n=88) in shSCN4B and shCTL, respectively, P<0.001, Students’ t-test; Fig. 7a). Using scanning electron microscopy (SEM), we also noticed a decreased number of filopodia-like structures per cell (medians were 19.0 (n=60) versus 44.5 (n=66) in shSCN4B and shCTL, respectively, P<0.001, Mann–Whitney rank sum test; Fig. 7b,c), and an increase in the number of blebs per cell (medians were 100.5 (n=82) versus 27.5 (n=82) in shSCN4B and shCTL, respectively, P<0.001 Mann–Whitney rank sum test; Fig. 7b–d). These blebs were relatively small with diameters between 0.5 and 0.7 μm (Fig. 7b). To confirm that the filopodia-like structures observed by SEM were real filopodia, and not retraction fibres, we also stained F-actin and performed confocal imaging (Fig. 7e). In agreement with results previously obtained in SEM, shSCN4B cells displayed less filopodia than shCTL cells (median were 8 (n=66 cells) versus 24.5 (n=46 cells) filopodia/cells for shSCN4B and shCTL, respectively, P<0.001, Mann–Whitney rank sum test; Fig. 7e,f). Furthermore, filopodia in shSCN4B cells were smaller than in shCTL cells (median length were 3.43 versus 5.61 μm in shSCN4B and shCTL, respectively, P<0.001, Mann–Whitney rank sum test; Fig. 7e–g). The morphological changes observed in shSCN4B cells might result from a transition towards an amoeboid phenotype that could confer cancer cells the ability to squeeze and migrate through small gaps of the ECM. Using SEM, we confirmed that shSCN4B cells demonstrated the ability to migrate through Matrigel, probably after they opened a narrow interstice by focalized degradation (Fig. 7h, Supplementary Fig. 7). The interconversions between mesenchymal and amoeboid modes of invasion are known to be orchestrated by RhoGTPases and the amoeboid movement mainly relies on the RhoA-ROCK-pMLCII signalling pathway. We therefore assessed the proportion of active (GTP-bound) RhoGTPases in shCTL and shSCN4B cells by pull-down assays (Fig. 7i) and found a significant increase in RhoA concomitant with decreases in Rac1 and Cdc-42 activities (Fig. 7j). Furthermore, the inhibition of myosin II with blebbistatin42 significantly reduced shSCN4B cancer cell invasiveness by approximately 27%, while it had no effect on shCTL cell invasiveness (Fig. 7k). Interestingly, proximity ligation assays indicated a close association of the SCN4B/β4 protein and RhoA in shCTL cells while no signal was observed in shSCN4B cells (Fig. 7l). Because the reduced expression of SCN4B increases cancer cell aggressiveness, we investigated whether its stable experimental overexpression (oeSCN4B) could have opposite effects. The overexpression of SCN4B gene, confirmed by qPCR and western blotting experiments (Supplementary Fig. 8a,b), significantly reduced cancer cell invasiveness by 51.6±6.8% (n=6) in oeSCN4B as compared with control cells and by about 82% as compared with shSCN4B cells (Fig. 8a). To further investigate the regulation of NaV activity and NaV-dependent invasiveness, we performed invasion experiments in the presence or absence of 30 μM TTX. While TTX reduced the invasion of control (oeCTL) cells to an extent similar to that found in wild-type or shCTL cells (i.e., a reduction by 30.7±4.0%, n=6), it had no further effect on reducing the invasive properties of oeSCN4B cells (Fig. 8b). Overexpressing SCN4B did not affect cancer cell growth and viability (Supplementary Fig. 8c). These results suggested that overexpression of SCN4B not only reduced the invasiveness via the SCN4B/β4 protein-dependent signalling pathway but also inhibited the NaV-dependent mesenchymal invasion. To test this hypothesis, we analysed INa in oeSCN4B cells. The maximal peak INa was higher in oeSCN4B cells as compared with oeCTL or shCTL cells, while the sensitivity to TTX remained unchanged (Fig. 8c and Supplementary Fig. 8d). In oeSCN4B cells, the INa activation–voltage relationship was not different from that of oeCTL cells (V1/2–activation were −42.8±0.4 mV (n=43) and −41.9±0.8 mV (n=15), respectively; Fig. 8d) and the availability–voltage relationship was slightly, but not significantly, shifted towards hyperpolarized values in oeSCN4B as compared with those obtained with oeCTL cells (V1/2–availability were −88.6±1.0 and −86.7±0.5 mV, respectively, P=0.459, Mann–Whitney rank sum test; Fig. 8d). This result was not statistically significant when compared with the V1/2–availability in shCTL cells (−86.3±1.6 mV, P=0.146, Mann–Whitney rank sum test), but was significant when compared with the V1/2–availability in shSCN4B cells (−80.6±1.6 mV, P<0.001, Mann–Whitney rank sum test). We however measured a decrease in the persistent INa current (Fig. 8e) and in INa persistent/peak ratio in oeSCN4B cells as compared with oeCTL (Fig. 8f) and a significant reduction of the focalized Matrigel degradation (Fig. 8g), suggesting that overexpression of SCN4B/β4 protein slightly reduced the persistent window current and its associated ECM proteolytic (mesenchymal) activity in cancer cells. This observation was also accompanied with reductions in the cell circularity index (from 0.49±0.02 (n=73) to 0.38±0.02 (n=73) in oeCTL and oeSCN4B, respectively, P<0.001, Student’s t-test; Fig. 8h and Supplementary Fig. 9a), in the migration speed (medians are 0.983 μm min−1 for oeCTL (n=47) and 0.520 μm min−1 for oeSCN4B cells (n=47), P<0.001, Mann–Whitney rank sum test; Fig. 7i) and in RhoA activity (Fig. 8j,k). Overall, these results indicate that SCN4B/β4 protein reduces mesenchymal invasion and prevents amoeboid migration by reducing NaV and RhoA activities, respectively.
SCNxB/β proteins, initially characterized as being ‘auxiliary’ subunits of NaV, have been proposed to also act as cell adhesion molecules (CAMs) via their immunoglobulin-like extracellular domain14. To investigate the participation of SCN4B/β4 protein domains in the phenotypical changes observed, we constructed different variants of the protein intended to be overexpressed in shSCN4B cells. For this purpose, the nucleotide sequence was mutated so that the transcripts would not to be targeted by the shRNA. Eight nucleotides were substituted without altering the amino acid sequence. The ‘full-length’ sequence allowed the expression of a wild-type β4 protein in shSCN4B cells. We also constructed two truncated variants: an N-terminus-truncated protein (from residue M1 to T161), called ‘ΔN-ter’, containing the transmembrane and C-terminus intracellular domains of the SCN4B/β4 protein but completely devoid of the Ig-like extracellular domain, and a SCN4B/β4 protein truncated in the C-terminus region, from residue K185, and identified as being ‘ΔC-ter’ (Fig. 9a). The ΔC-ter construct contained the same eight substituted nucleotides with conservation of the amino acid sequence so that it could be expressed in shSCN4B cells. The protein expression level of these variants at the plasma membrane was similar (Supplementary Fig. 9b). The reintroduction of the full-length SCN4B/β4 protein significantly reduced cancer cell invasiveness as compared with the empty vector (pSec), and as such, performed as an effective rescue. Interestingly, the ΔC-ter variant, which possessed the extracellular domain, was completely ineffective, whereas the ΔN-ter variant reduced cell invasiveness to the same extent as the full-length protein (Fig. 9b). The invasiveness of shSCN4B cells overexpressing the ΔN-ter variant or the full-length protein was still superior to the one of shCTL cells by about 50%. In agreement with these results, only the full-length and the ΔN-ter proteins reduced the speed of migration and RhoA activity (Fig. 9c,d) in shSCN4B cells, therefore demonstrating that the intracellular C-terminus of the SCN4B/β4 protein, and not the extracellular Ig-like domain, is needed to inhibit invasiveness in breast cancer cells.
Two important steps in tumour progression are the entry of cancer cells into the vasculature and their dissemination into the general circulation before colonization of secondary organs. This progression requires cellular intravasation/extravasation abilities that can be assessed in vitro using transendothelial invasion (with an initial migration through the ECM, before crossing over the endothelium) and migration protocols, respectively. In both experiments, the downregulation of SCN4B importantly potentiated the capacity of cancer cells to migrate through a monolayer of endothelial cells, after having invaded (invasion), or not (migration), a layer of matrigel, as compared with shCTL or oeSCN4B cells (Fig. 10a,b). We then studied the involvement of the SCN4B gene and its expression product, the SCN4B/β4 protein, in tumour progression. For this we used oeSCN4B and shSCN4B cells as models of cancer cells, coming from the same lineage and solely differing by the presence of high levels of SCN4B/β4 protein (similar to non-cancerous mammary cells, oeSCN4B cells) or the complete absence of SCN4B/β4 protein (similar to high-grade cancers, shSCN4B cells). We developed two in vivo models in immunodepressed mice. In the first model, we assessed the importance of SCN4B expression in human breast cancer cells for the colonization of organs. ShSCN4B or oeSCN4B cells, both having identical in vitro growth and viability properties and similarly expressing the luciferase gene (Supplementary Fig. 8c–e), were injected in the tail vein of NMRI nude mice. After 9 weeks, there was no statistical difference in animal body weights between the two experimental groups (Supplementary Fig. 8f). Mice were euthanized and the isolated organs (lungs, brain, liver, bones from spine/ribs and legs) were analysed ex vivo for bioluminescent imaging after luciferin injection (Fig. 10c). In the shSCN4B group, all mice showed lung metastases (7/7), while in the oeSCN4B group, only one mouse out of eight had lung colonization and there was no bioluminescent signal in other organs (Yate’s χ2, P=0.0041). Altogether, there was a strong reduction in lung colonization by cancer cells overexpressing SCN4B/β4 compared with non-expressors (Fig. 10d). In the other model of orthotopic xenograft of mammary cancer, in which shSCN4B or oeSCN4B cells were injected into the mammary fat pad of NOD SCID mice, the primary tumour growth was analysed as a function of time for 23 weeks. There was no statistical difference in the evolution of animal body weights between the two experimental groups (8 mice/group; Supplementary Fig. 10a). The growth of primary mammary tumours, measured with a calliper (Supplementary Fig. 10b) or by bioluminescent imaging (Fig. 10e,f), was reduced in mice implanted with oeSCN4B cells compared with those implanted with shSCN4B. At the end of the study, primary mammary tumours from the two groups were analysed and the non-expression of SCN4B/β4 in the shSCN4B group or expression in the oeSCN4B group was confirmed by immunohistochemistry (Fig. 10g). Finally, 5/8 mice implanted with shSCN4B cells had more metastatic foci (3/8 in lungs, 1 in spine and 1 in posterior legs) compared with those implanted with oeSCN4B cells (only 1/8 mouse showed lung metastases) (Yates χ2 P-value=0.039; Supplementary Fig. 10c–e). The presence of human breast cancer cells in mouse lungs was confirmed by human cytokeratin 7 immunoreactivity (Fig. 10h). All together, these results indicate that the overexpression of SCN4B in cancer cells reduces primary tumour growth and metastatic colonization.
SCNxB/β proteins were initially isolated from rat brain along with pore-forming NaV sodium channels43. They all exhibit an extracellular N-terminus containing an immunoglobulin domain. With the exception of the soluble β1B protein44, all SCNxB/β subunits have a single α-helical transmembrane domain and a short intracellular domain45. SCNxB/β proteins were demonstrated to interact, through covalent or non-covalent associations, with pore-forming Nav channels46,47,48,49, to regulate their trafficking to the plasma membrane, as well as their biophysical50 and pharmacological51,52,53 properties. Because of these functions, SCNxB/β proteins were initially characterized as being auxiliary subunits of pore-forming Nav in excitable cells. Besides these canonical roles, it has later been proposed that they might also have other specific cellular functions54. Indeed, the presence of an Ig motif in their extracellular domain, similar to that of CAMs such as integrins, cadherins and selectins, argued for a possible function in cell adhesion properties55,56. SCN1B/β1 and SCN2B/β2 subunits were demonstrated to form both trans-homophilic and trans-heterophilic cell–cell and cell–matrix adhesions in cells expressing Nav, such as neurons in which they were proven to be critical for neurites outgrowth, axonal fasciculation and interactions with glial cells14.
The expression of SCNxB/β proteins and their physiological role in non-excitable cells have not been characterized. Their participation in oncogenic processes was not studied as much as that of pore-forming Nav proteins17. The most studied isoform in cancer is SCN1B/β1, and results published so far could appear contradictory. An initial in vitro study indicated that the expression of SCN1B/β1 was higher in poorly invasive than in highly invasive breast cancer cells. In weakly invasive MCF-7 cells, the downregulation of SCN1B/β1 reduced cell adhesion, and increased cell migration. Correlatively, its overexpression in MDA-MB-231 cells, increased cell process length and adhesion, reduced lateral motility and cell proliferation33. These results suggested that SCN1B/β1 expression could prevent cancer cell invasion. In contrast, a recent study demonstrated that SCN1B/β1 was overexpressed in breast cancer biopsies, compared with non-cancer breast samples. Furthermore, the overexpression of SCN1B/β1 in breast cancer cells potentiated their invasiveness in vitro, and increased primary tumour growth and metastatic development in vivo27. The proinvasive role of SCN1B/β1 was proposed to depend on both Nav-dependent and Nav-independent mechanisms, relying on the extracellular Ig domain of the protein27. In this study, we found that SCN1B gene is downregulated in breast cancer compared with normal breast tissues. Nevertheless, we identified SCN1B as being the SCNxB gene the most highly expressed in cancer cells. Interestingly, the downregulation of SCN1B/β1 reduces in vitro invasiveness of MDA-MB-231, which is in line with results obtained by Nelson et al.27 showing this isoform displays pro-invasive roles in vitro. Concerning the SCN4B/β4 protein, there was no information regarding its potential involvement in the oncogenic process. Only one study identified that SCN4B expression was decreased in cervical cancer biopsies compared with non-cancer samples57.
Aggressive cancer cells present an important plasticity enabling them to use different invasion modes, conferring an ability to adapt to their microenvironment, matrix composition, meshwork, eventually promoting tumour growth and metastases development. While RhoGTPases are involved in invasive phenotypes, the upstream mechanism regulating their respective activation is not known. Here, we identified the SCN4B gene as an important modulator of RhoA activity and invasiveness in cancer cells. The SCN4B/β4 protein was known to be an auxiliary subunit of NaV in excitable cells. The present study shows for the first time that this protein is expressed in non-cancer epithelial cells, which do not express NaV, suggesting a physiological non-auxiliary function. Furthermore, the expression of SCN4B/β4 is reduced in cancer tissues, and more specifically when tumours gain invasive properties (transition from grade I to grade II). SCN4B/β4 is almost absent in high-grade tumours and metastases. At the cellular level, the loss of SCN4B/β4 in cancer cells further stimulates their invasiveness by favouring amoeboid migration, supported by the overactivation of the RhoA pathway, yet keeping the ECM-degradative activity intact. Conversely, the overexpression of SCN4B/β4 reduces cancer cell invasiveness, tumour growth and metastatic progression, supporting our proposal that the SCN4B gene is a metastasis-suppressor gene.
This study also demonstrates that the SCN4B/β4 protein acts both as NaV pore-subunit regulator and as RhoGTPases regulator in cancer cells. Indeed, the suppression of SCN4B gene expression in cancer cells reduced the peak sodium current due to NaV1.5 activity. This could be attributed to a reduction of SCN5A expression, and to a consequent reduction of channel density at the plasma membrane. However, we found that channels expressed at the plasma membrane of cancer cells show a higher activity through an increased persistent window current. The loss of interaction between NaV1.5 and SCN4B/β4 protein might therefore be responsible for a gain-of-function of NaV1.5. This is similar to what was observed in the case of the cardiac long QT syndrome LQT10, in which a missense mutation in the SCN4B gene, resulting in the L1759F amino acid substitution in the β4 protein, induces a gain-of-function of NaV1.5 with an increased persistent sodium current29. Because of this regulation of NaV1.5, the loss of SCN4B expression in cancer cells left NaV-dependent mesenchymal invasiveness unaffected, through the maintenance of a persistent sodium current that regulates ECM proteolysis, but also increased NaV-independent amoeboid-related cell migration, through the dynamic regulation of RhoGTPases. This latter property is under the control of the intracellular C-terminus and is independent of the extracellular CAM domain. These results contrast with those obtained with the pro-invasive SCN1B/β1 for which the extracellular Ig domain is crucial for its CAM function in breast cancer cells27. Our results further indicate that the intracellular C-terminus of SCN4B/β4 might play a critical role in maintaining epithelial function and integrity.
In conclusion, this study shows that the loss of SCN4B gene expression in cancer cells promotes the acquisition of an amoeboid–mesenchymal hybrid phenotype, while its overexpression reduces both mesenchymal and amoeboid invasive capacities. The expression level of SCN4B could therefore represent an important prognosis marker in cancers from epithelial origin.
Prognostic analyses of gene expression in breast cancers
Analyses were performed using the software Breast Cancer Gene-Expression Miner v3.0(bc-GenExMiner v3.0; http://bcgenex.centregauducheau.fr) developed by the Integrated Center of Oncology René Gauducheau (Nantes-Saint Herblain, France), based on DNA microarrays results collected from published cohorts58. Briefly, several statistical tests were conducted on each cohort and on all cohorts pooled together with data from all studies previously converted to a common scale with a suitable normalization. The prognostic impact of each gene was evaluated by means of the univariate Cox proportional hazards model. Results were displayed by cohorts and pools and are illustrated in a forest plot. Kaplan–Meier curves were then obtained on the pool with the gene values dichotomized according to gene expression median (calculated from the pool). Cox results corresponding to dichotomized values were displayed on the graph.
In silico RNA expression analyses
Expression of SCNxB genes in cancer tissues was studied using the web-based ‘The Cancer Genome Atlas’ (http://cancergenome.nih.gov) from the US National Cancer Institute and the National Human Genome Research Institute that integrates RNE sequences databases. Data are expressed as RPKM (Reads Per Kilobase per Million).
The degree of SCN4B/β4 protein expression in normal and dysplastic mammary tissues, as well as in mammary ductal and lobular carcinomas, was analysed by standard immunohistochemistry procedures. Tissue microarrays (TMA) from formalin-fixed, paraffin-embedded mammary tissues were purchased from US Biomax Inc. (ref. BR1003, BC081120, BR10010a; Rockville, USA), comprising normal, hyperplastic and dysplastic mammary samples, lobular and ductal mammary carcinomas (separated in well (grade I), moderately (grade II) and poorly (grade III) differentiated carcinomas) and LNM samples. Similar analyses were also performed in a lung TMA (LC951; US Biomax Inc) containing normal lung, cancer lung (grades IA to IIIB) tissues and metastases (in lymph nodes, bones and intestine). TMA were analysed in blind conditions by C.M.-C. (anatomopathologist) and P.P. (immunologist, Clinical University Hospital Virgen de la Arrixaca, Murcia, Spain). Normal breast and cancer biopsies from the University-Hospital of Tours were coming from the tumour collection declared to the French Ministry of Research (No. DC2008-308) and were prepared by R.G. and analysed by G.F. (clinical anatomothologist at the University-Hospital of Tours, France).
Briefly, after deparaffinization and rehydration, sections were treated with a high-pH (Tris buffer/EDTA, pH 9.0) target retrieval procedure (Dako PT-link; Dako, USA). Endogenous peroxidase was then blocked by a commercial solution (Dako REAL), and incubated overnight with a 1/100 dilution of the primary polyclonal rabbit antibody anti-SCN4B/β4 protein (HPA017293; Sigma-Aldrich, France) at 4 °C. The antibody used recognizes the following extracellular peptide sequence of the β4 protein: LLPCTFSSCFGFEDLHFRWTYNSSDAFKILIEGTVKNEKSDPKVTLKDDDRITLVGSTKEKMNNISIVLRDLEFSDTGKYTCHVKNPKENNLQHHATIFLQVV. Sections were then incubated with a commercial anti-rabbit-labelled polymer (Dako EnVision FLEX; Dako) for 30 min at RT. Immunoreaction was finally revealed with 3-3′ diaminobenzidine solution (Dako) for 5 min. Positive reaction was identified by a cytoplasmic dark-brown precipitate. To determine the degree of protein expression in tissues, a qualitative scale was used, for negative (−), weak (+) and strong (++) cytoplasmic expression. A unique score was given per core. The number of samples is representative of the number of cores, which is equivalent to the number of patients.
Lung and primary tumours from in vivo mouse experiments were fixed in formalin, included in paraffin, and cut in 5 μm tissue sections. Slides were deparaffinized, rehydrated and heated in citrate buffer pH 6 for antigenic retrieval. The primary antibodies included monoclonal anti-human cytokeratin 7 (clone OV-TL 12/30; Dakocytomation, Glostrup, Denmark, dilution 1/100, 1 h), used on lung sections, and polyclonal anti-mouse SCN4B/β4 (dilution 1/100, 1 h), used on primary tumour sections. Immunohistochemistry was performed using the streptavidin-biotin-peroxidase method with diaminobenzidine as the chromogen (Kit LSAB, Dakocytomation). Slides were finally counterstained with haematoxylin. Negative controls were obtained after omission of the primary antibody or incubation with an irrelevant antibody.
Reagents and antibodies
TTX was purchased from Latoxan (France), and A803467 from R&D systems (France). Fluorescent probes and conjugated antibodies were purchased from Fisher Scientific (France). Drugs and chemicals were purchased from Sigma-Aldrich.
Cell culture and cell lines
All cell lines were from the American Type Culture Collection (LGC Promochem, France) and were grown at 37 °C in a humidified 5% CO2 incubator. The immortalized normal mammary epithelial cells MCF-10A were cultured in Dulbecco's modified Eagle’s medium (DMEM)/Ham’s F-12, 1:1 mix containing 5% horse serum (Invitrogen, France), 10 μg ml−1 insulin, 20 ng ml−1 epidermal growth factor, 0.5 μg ml−1 hydrocortisone and 100 ng ml−1 cholera toxin. MCF-7, MDA-MB-468 and MDA-MB-435s breast cancer cells were cultured in DMEM supplemented with 5% fetal calf serum (FCS). PC3 prostate, H460 and A549 non-small-cell lung cancer cells were cultured in DMEM supplemented with 10% FCS. MDA-MB-231-Luc human breast cancer cells, stably expressing the luciferase gene22, were cultured in DMEM supplemented with 5% FCS. We constructed a lentiviral vector encoding a short hairpin RNA (shRNA) specifically targeting human SCN5A transcripts35. The sequence encoding shSCN5A, inhibiting the expression of NaV1.5 protein, was obtained by DNA polymerase fill-in of two partially complementary primers: 5′-GGATCCCCAAGGCACAAGTGCGTGCGCAATTCAAGAGA-3′ and 5′-AAGCTTAAAAAAAGGCACAAGTGCGTGCGCAATCTCTTGAA-3′.
Similarly, we constructed a lentiviral vector encoding a short hairpin RNA (shRNA) specifically targeting human SCN4B transcripts, inhibiting the expression of β4 protein. The sequence encoding the shSCN4B was obtained by DNA polymerase fill-in of two partially complementary primers; this method also allowed the introduction of two restriction enzyme sites to facilitate manipulations. Forward primer: shβ4-BamHI 5′-GGATCCCCCAGCAGTGACGCATTCAAGATTCTTCAAGAGA-3′and reverse primer: shβ4-HindIII 5′-AAGCTTAAAAACAGCAGTGACGCATTCAAGATTCTCTCTTGAA-3′. We also constructed a lentiviral vector expressing a null-target shRNA (pLenti-shCTL), using the following primers: 5′-GGATCCCCGCCGACCAATTCACGGCCGTTCAAGAGACG-3′ and 5′-AAGCTTAAAAAGCCGACCAATTCACGGCCGTCTCTTGAACG-3′. We constructed an expression plasmid containing the human SCN4B coding sequence to overexpress β4. This sequence was synthetized by Proteogenix (France) and inserted in pcDNA3.1, using the following sequences: 5′-GGATCCGCCGCCACC-3′ and 5′-GCGGCCGCCTCGAG-3′. We designed the mutated sequences coding for the ‘Full-length SCN4B/β4′ and truncated proteins, which were then synthetized by Proteogenix (France) and all sequences obtained were inserted into pSecTag2 hygro B vector (V910-20; Invitrogen) with the In-fusion HD cloning Plus kit (Clonetech). We constructed a plasmid containing the sequence coding for the N-terminally truncated (from residue 1 to residue T161, ‘ΔN-ter’) protein containing the transmembrane and C-terminal intracellular domain of the SCN4B/β4 protein. The sequence was obtained by PCR elongation using two specific primers: forward primer 5′-GCGCCGTACGAAGCTGACCTGGAGTTCAGCGAC-3′ and reverse primer 5′- ACACTGGAGTGGATCTCACACTTTTGAAGGTGGTT-3′. Similarly, an SCN4B/β4 protein truncated in the C-terminus, starting with residue K185 and identified as being ‘ΔC-ter’ was designed. Importantly, for ‘ΔC-ter’ and ‘full-length’ the nucleotide sequence was mutated by substitution to avoid targeting by the shRNA targeting native SCN4B gene. The protein sequence remained unaffected. The sequence targeted by the shRNA was 5′-CAGCAGTGACGCATTCAAATTC-3′, while the substituted untargeted sequence was 5′-TAGTAGCGATGCCTTTAAATAC-3′. All variants exhibited an His-Tag and their expression and subcellular localization, after transfection of shSCN4B cells, were assessed by epifluorescence imaging using a primary rabbit anti-HisTag antibody (T2767; Invitrogen), and a secondary goat anti-rabbit Texas Red antibody (Thermofisher, France).
Mycoplasma contamination tests were performed every week (Lonza, MycoAlert Mycoplasma Detection Kit).
Reverse transcription of RNA and real-time PCR
Total RNA extraction was performed using the RNAgents Total RNA Isolation System (Promega, France). RNA yield and purity were determined by spectrophotometry and only samples with an A260/A280 ratio above 1.6 were kept for further experiments. Total RNA were reverse-transcribed with the RT kits Ready-to-go You-prime First-Strand Beads (Amersham Biosciences, UK). Random hexamers pd(N)6 5′-phosphate (0.2 μg; Amersham Biosciences) were added and the reaction mixture was incubated at 37 °C for 60 min. Real-time PCR experiments were performed as previously described24. Primers sequences are given in Supplementary Table 1.
siRNA transfection and efficacy assessment
MDA-MB-231-Luc human breast cancer cells were transfected with 20 nM siRNA targeting the expression of SCN1B (siSCN1B, sc-97849), SCN2B (siSCN2B, sc-96252), SCN4B (siSCN4B, sc-62982) or scrambled siRNA (siCTL, siRNA-A sc-37007), which were produced by Santa Cruz Biotechnology and were purchased from Tebu-Bio (France). Transfection was performed with Lipofectamine RNAi max (Invitrogen) according to the manufacturer’s instructions, and used 24 h after transfection. The efficiency of siRNA transfection was verified by qPCR using an iCycler system (BioRad, USA) and by western blotting.
Western blotting experiments
Cells were washed with phosphate-buffered saline (PBS) and lysed in the presence of a lysis buffer (50 mM Tris, pH 7, 100 mM NaCl, 5 mM MgCl2, 10% glycerol, 1 mM EDTA), containing 1% Triton-X-100 and protease inhibitors (Sigma-Aldrich). Cell lysates were cleared by centrifugation at 16,000 g for 10 min. Western blotting experiments were performed according to standard protocols. Total protein concentrations were determined using the Pierce BCA Protein Assay Kit Thermoscientific (Fisher Scientific, France). Protein sample buffer was added and the samples were boiled at 100 °C for 3 min. Total protein samples were electrophoretically separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis in 10% gels, and then transferred to polyvinylidene fluoride membranes (Millipore, USA). The SCNxB/β proteins were detected using anti-SCN1B/β1 (1/1,000, AV35028; Sigma-Aldrich), anti-SCN2B/β2 (1/200, HPA012585; Sigma-Aldrich) and anti-SCN4B/β4 (1/1,000, HPA01293, Sigma-Aldrich) rabbit polyclonal primary antibodies and horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody at 1:2,000 (TebuBio, France). HSC70 protein was detected as a sample loading control using anti-HSC70 mouse primary antibody at 1:30,000 (TebuBio) and horseradish peroxidase -conjugated anti-mouse-IgG secondary antibodies at 1:2,000 (TebuBio). Proteins were revealed using electrochemiluminescence-plus kit (Pierce ECL Western Blotting Substrate; Fisher Scientific) and captured on Kodak Bio-Mark MS films (Sigma-Aldrich). Full scans of western blots are shown in Supplementary Fig. 11.
RhoGTPases pull-down assays
Pull-down assays were performed according to the manufacturer’s protocol (Cat#BK030, RhoA/Rac1/Cdc42 Activation Assay Combo Biochem Kit; Cytoskeleton, Inc.). Briefly, cells were washed on ice with ice-cold PBS, then lysed and scraped with ice-cold cell lysis buffer containing protease inhibitors (Sigma-Aldrich). Cell lysates were clarified by centrifugation at 10,000 g, +4 °C, 1 min, and the supernatant was snap-frozen in liquid nitrogen. Ten microlitres of clarified cell lysate were used to perform the protein assay (Cat#23225, BCA protein assay; Thermofisher). Total proteins (300 μg) were incubated with 10 μg of PAK-PDB beads or 30 μg Rhotekin-RBD beads for 1 h on a rotator at +4 °C. Samples and beads were washed with 500 μl washing buffer, resuspended in Laemmli buffer and boiled for 2 min prior to performing western blotting experiments. Antibodies for RhoA, Rac1 and Cdc42 were provided with the kit and used according to the manufacturer’s protocol.
Patch pipettes were pulled from borosilicate glass to a resistance of 3–5 MΩ. Currents were recorded, in whole-cell configuration, under voltage-clamp mode of the patch-clamp technique, at room temperature, using an Axopatch 200B patch clamp amplifier (Axon Instrument, USA). Analogue signals were filtered at 5 kHz, and sampled at 10 kHz using a 1,440A Digidata converter. Cell capacitance and series resistance were electronically compensated by about 60%. The P/2 sub-pulse correction of cell leakage and capacitance was used to study Na+ current (INa). Sodium currents were recorded by depolarizing the cells from a holding potential of −100 mV to a maximal test pulse of −30 mV for 30 ms every 500 ms. The protocol used to build sodium current–voltage (INa–V) relationships was as follows: from a holding potential of −100 mV, the membrane was stepped to potentials from −80 to +60 mV, with 5-mV increments, for 50 ms at a frequency of 2 Hz. Availability–voltage relationships were obtained by applying 50 ms prepulses using the INa–V curve procedure followed by a depolarizing pulse to −5 mV for 50 ms. In this case, currents were normalized to the amplitude of the test current without a prepulse. Conductance through Na+ channels (gNa) was calculated as already described21. Current amplitudes were normalized to cell capacitance and expressed as current density (pA/pF). The Physiological Saline Solution had the following composition (in mM): NaCl 140, KCl 4, MgCl2 1, CaCl2 2, D-glucose 11.1 and HEPES 10, adjusted to pH 7.4 with NaOH (1 M). The intrapipette solution had the following composition (in mM): KCl 130, NaCl 15, CaCl2 0.37, MgCl2 1, Mg-ATP 1, EGTA 1, HEPES 10, adjusted to pH 7.2 with KOH (1 M).
Measurement of intracellular pH
Cells were incubated for 30 min at 37 °C in Hank’s medium containing 2 μM BCECF-AM (2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein; excitation 503/440 nm; emission 530 nm). Excess dye was removed by rinsing the cells twice with Physiological Saline Solution. H+ efflux was measured as previously described25,59.
Cells were seeded at 4 × 104 cells per well in a 24-well plate and were grown for a total of 5 days. Culture media were changed every day. Viable cell numbers were assessed by the tetrazolium salt assay as previously described24 and normalized to the appropriate control condition (MDA-MB-231-Luc or shCTL).
In vitro invasion assays
Cell invasiveness was analysed as previously described25 using culture inserts with 8-μm pore size filters covered with Matrigel (Becton Dickinson, France). Briefly, the upper compartment was seeded with 6 × 104 cells in DMEM supplemented with 5% FCS. The lower compartment was filled with DMEM supplemented with 10% FCS (or 15% FCS for non-small-cell lung and prostate cancer cells), as a chemoattractant. After 24 h at 37 °C, remaining cells were removed from the upper side of the membrane. Cells that had invaded and migrated through the insert and were attached to the lower side were stained with DAPI and counted on the whole area of the insert membrane. In vitro invasion assays were performed in triplicate in each separate experiment.
3D invasion assays
Five thousand shCTL or shSCN4B MDA-MB-231-luc cells were suspended in DMEM containing Matrigel (Corning; ref 356230, 2.7 mg ml−1 final concentration), seeded in 96-well microplates and placed at 37 °C, 5% CO2. One hundred microlitres of DMEM+5% FBS were added in each well 30 min after seeding. Time-lapse acquisitions (1 image per 30 min for a total duration of 48 h) were performed in the presence or absence of the MMP inhibitor GM6001 (10 μM final concentration) to monitor 3D cancer cell invasiveness and started 4 h after cell seeding.
For the assessment of ECM degradation, cells were cultured for 24 h on glass coverslips coated with a matrix composed of Matrigel (4 mg ml−1, final concentration) and containing or not DQ-gelatin (25 μg ml−1, Thermofisher) as a fluorogenic substrate of gelatinases. They were then washed in PBS, fixed with 3.7% ice-cold paraformaldehyde in PBS. Cells were permeabilized with a solution containing 50 mM NH4Cl, 1% BSA and 0.02% saponin, then saturated in a solution containing 3% BSA and 3% normal goat serum . F-actin was stained with phalloidin-AlexaFluor594. Epifluorescence microscopy was performed with a Nikon TI-S microscope and analysed using both NIS-BR software (Nikon, France) and ImageJ software 1.38I (http://rsbweb.nih.gov/ij). Pixels corresponding to the co-localization of F-actin condensation areas and focal spots of DQ-gelatin proteolysis (excitation wavelength: 495 nm, emission wavelength: 515 nm) were quantified per cell, giving a Matrix-focalized-degradation index.
Phospho (Y421)-cortactin was detected using the primary rabbit anti-pY421 cortactin antibody (Millipore) and the fluorescent-conjugated secondary anti-rabbit TexasRed antibody (Thermofisher). Pixels corresponding to the co-localization of phospho-cortactin and focal spots of DQ-gelatin proteolysis were quantified per cell using ImageJ software 1.38I.
For the assessment of cell morphology, cells were cultured for 24 h on glass coverslips and F-actin was stained with phalloidin-AlexaFluor594. A circularity index was calculated from pictures as being 4π·area/perimeter2. A value approaching 0 indicates an increasingly elongated shape while a value of 1.0 indicates a perfect circle.
5,000 cells were seeded on Labtek coverslips (Thermofisher) and placed in the incubator at 37 °C, 5% CO2 for 24 h. They were then washed twice in PBS, fixed with 3.7% ice-cold paraformaldehyde in PBS. Cells were permeabilized with a solution containing 50 mM NH4Cl, 1% BSA and 0.02% saponin, then saturated in a solution containing 3% BSA and 3% normal goat serum. F-actin was stained with phalloidin-AlexaFluor488 (Thermofisher). Cells were observed on a confocal microscope, at × 600 magnification (Olympus Fluoview FV500 Laser Scanning Confocal Biological Microscope) and image acquisition was performed using Fluoview 500 v.5 software (Olympus, Tokyo, Japan).
Scanning electron microscopy
Cells were fixed by incubation for 24 h in 4% paraformaldehyde, 1% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2). They were then washed in PBS and post-fixed by incubation with 2% osmium tetroxide for 1 h. Samples were then fully dehydrated in a graded series of ethanol solutions and dried in hexamethyldisilazane (HMDS; Sigma, USA). Finally, cells were coated with 40 Å platinum, using a GATAN PECS 682 apparatus (Pleasanton, USA), before observation under a Zeiss Ultra plus FEG-SEM scanning electron microscope (Oberkochen, Germany). Cells were pseudocolored using PowerPoint software (Microsoft, USA).
Zebrafish invasion assays
Zebrafish (Danio rerio), from the Zebrafish International Resource Centre (ZIRC), were maintained in re-circulating tanks according to the standard procedures61. Adult fishes were maintained at 26 °C, with a light/dark cycle of 14/10 h, and were fed twice daily, once with dry flake food (PRODAC) and once with live Artemia salina (MC 450, IVE AQUACULTURE). Zebrafish embryos were maintained in egg water at 28.5 °C, fed for 5 days with NOVO TOM and with live A. salina at 11 days of life. All experiments were performed in compliance with the Guidelines of the European Union Council for animal experimentation (86/609/EU) and were approved by the Bioethical Committee of the University Hospital Virgen de la Arrixaca (Spain). The colonization of zebrafish embryos was previously described35. Briefly, MDA-MB-231 breast cancer cells transfected with siRNA targeting the expression of SCN4B gene (SiSCN4B) or null-target siRNA (siCTL) were trypsinized 24 h after transfection, washed and stained with the vital cell tracker red fluorescent CM-Dil (Vibrant; Invitrogen). Fifty labelled cells were injected into the yolk sac of dechorionated zebrafish embryos using a manual injector (Narishige). Fish with fluorescently labelled cells appearing outside of the implantation area at 2 h post-injection were excluded from further analysis. All other fishes were incubated at 35 °C for 48 h and analysed with a SteReo Lumar V12 stereomicroscope equipped with an AxioCam MR5 camera (Carl Zeiss). The evaluation criteria for embryos being colonized by human cancer cells was the presence of more than five cells outside of the yolk sac. A zebrafish (ZF) colonization index was calculated as being the proportion of embryos being colonized (by at least five human cancer cells) in the siSCN4B condition divided by the proportion of invaded embryos in the siCTL condition.
Human umbilical vein endothelial cell (HUVEC; Promocell, Germany) were cultured in medium 199 (1X; Gibco) containing 20% FCS, 100 μg ml−1 heparin and 50 μg ml−1 of endothelial cell growth supplement (Sigma-Aldrich) and were used up to the fifth passage. For transendothelial migration, 100,000 HUVEC were seeded in the upper side of 8-μm-pore sized transwell migration inserts (Corning, France) and grown for 6 days. Endothelial culture medium was carefully removed and replaced by DMEM 10% FCS in the lower chamber and DMEM 5% FCS containing 60,000 MDA-MB-231 in the upper chamber. MDA-MB-231 cells were stained with 4 ng μl−1 CM-Dil (Thermo Fisher Scientific) prior seeding.
For transendothelial invasion, matrigel-coated transwell invasion inserts were inverted in order to plate the bottom surface with 100,000 HUVEC for 4 h at 37 °C, 5% CO2. Invasion chambers were then inverted again, inserted into a 24-well plates and HUVEC were cultured for 6 days in endothelial culture medium before plating MDA-MB-231 cells in the upper chamber. MDA-MB-231 cells were CM-Dil stained as described above. After 18 h of culture, transwells were washed in PBS and fixed in formaldehyde 4% for 20 min.
Transendothelial electrical resistance measurements were recorded every day for 6 days until the transendothelial resistance was stable (>10 kΩ) on the same transwells using an EVOM2 epithelial Voltohmmeter (World Precision Instruments, France).
In vivo tumour models
All animals were bred and housed at the In Vivo platform of the Cancéropôle Grand Ouest at Inserm U892 (Nantes, France) under the animal care license no. 44278. The project was approved by the French national ethical committee (APAFIS 00085.01).
Unanaesthetized six-week-old female NMRI Nude Mice (Charles River Laboratories) were placed into a plastic restraining device, and 2 × 106 MDA-MB-231-Luc cells (shSCN4B, 7 mice/oeSCN4B, 8 mice) suspended in 100 μl PBS were injected into the lateral tail vein through a 25-gauge needle as previously described22. At necropsy, ex vivo BLI measurement for each collected organ was performed within 15 min of D-luciferin intraperitoneal injection (150 mg kg−1). Photons emitted by cancer cells were counted by bioluminescent imaging (ΦimagerTM; BIOSPACE Lab) and expressed in counts per minute (c.p.m.).
Six-week-old female Rag2−/− Il2rg−/− mice (NOD SCID; Charles River Laboratories) were injected into the sixth right inguinal mammary fat pad with 2 × 106 MDA-MB-231-Luc cells (shSCN4B or oeSCN4B, 8 mice per group) suspended in 100 μl PBS while under isoflurane anaesthesia. Tumour growth was monitored by bioluminescence imaging22. Animal weight was measured every week or every other week, and the primary tumour volume (mm3) was measured with a calliper and calculated as length × height × width (in mm). Mice were euthanized 24 weeks following implantation of tumour cells and metastatic bioluminescence was measured22.
Statistical analyses on immunohistochemistry staining were performed using the Yate’s χ2 test using the online interactive Chi-square test software Quantpsy (http://www.quantpsy.org/chisq/chisq.htm). Other data were displayed as mean±s.e.m. and n=sample size, and were analysed using parametric statistical tests (Student’s t-test or ANOVA) when they followed a normal distribution and equal variances. Alternatively, when samples did not follow a normal distribution, or when variances failed to be comparable, data were displayed as box plots indicating the first quartile, the median and the third quartile, and squares for comparison of means. In these cases, adequate non-parametric statistical tests were used (Mann–Whitney rank sum tests, Dunn’s tests, ANOVA on ranks). Statistical analyses were performed using SigmaStat 3.0 software (Systat software Inc.) and statistical significance is indicated as *P<0.05; **P<0.01 and ***P<0.001. NS stands for not statistically different.
Expression of SCNxB genes in cancer tissues was studied using the web-based ‘The Cancer Genome Atlas’ (http://cancergenome.nih.gov). The authors declare that all other data supporting the findings of this study are available within the paper and its Supplementary Information files or available from the authors upon request.
How to cite this article: Bon, E. et al. SCN4B acts as a metastasis-suppressor gene preventing hyperactivation of cell migration in breast cancer. Nat. Commun. 7, 13648 doi: 10.1038/ncomms13648 (2016).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Parkin, D. M., Bray, F., Ferlay, J. & Pisani, P. Global cancer statistics, 2002. CA Cancer J. Clin. 55, 74–108 (2005).
Fidler, I. J. Understanding bone metastases: the key to the effective treatment of prostate cancer. Clin. Adv. Hematol. Oncol. 1, 278–279 (2003).
Friedl, P. & Alexander, S. Cancer invasion and the microenvironment: plasticity and reciprocity. Cell 147, 992–1009 (2011).
Linder, S., Wiesner, C. & Himmel, M. Degrading devices: invadosomes in proteolytic cell invasion. Annu. Rev. Cell Dev. Biol. 27, 185–211 (2011).
Brisson, L., Reshkin, S. J., Gore, J. & Roger, S. pH regulators in invadosomal functioning: proton delivery for matrix tasting. Eur. J. Cell Biol. 91, 847–860 (2012).
Wolf, K. & Friedl, P. Extracellular matrix determinants of proteolytic and non-proteolytic cell migration. Trends Cell Biol. 21, 736–744 (2011).
Sanz-Moreno, V. et al. Rac activation and inactivation control plasticity of tumor cell movement. Cell 135, 510–523 (2008).
Sahai, E. & Marshall, C. J. Differing modes of tumour cell invasion have distinct requirements for Rho/ROCK signalling and extracellular proteolysis. Nat. Cell Biol. 5, 711–719 (2003).
Wyckoff, J. B., Pinner, S. E., Gschmeissner, S., Condeelis, J. S. & Sahai, E. ROCK- and myosin-dependent matrix deformation enables protease-independent tumor-cell invasion in vivo. Curr. Biol. 16, 1515–1523 (2006).
Friedl, P. & Wolf, K. Tumour-cell invasion and migration: diversity and escape mechanisms. Nat. Rev. Cancer 3, 362–374 (2003).
Wolf, K. et al. Compensation mechanism in tumor cell migration: mesenchymal-amoeboid transition after blocking of pericellular proteolysis. J. Cell Biol. 160, 267–277 (2003).
Goldin, A. L. et al. Nomenclature of voltage-gated sodium channels. Neuron 28, 365–368 (2000).
Catterall, W. A. Voltage-gated sodium channels at 60: structure, function, and pathophysiology. J. Physiol. 590, 2577–2589 (2012).
O'Malley, H. A. & Isom, L. L. Sodium channel beta subunits: emerging targets in channelopathies. Annu. Rev. Physiol. 77, 481–504 (2015).
Black, J. A. & Waxman, S. G. Noncanonical roles of voltage-gated sodium channels. Neuron 80, 280–291 (2013).
Besson, P. et al. How do voltage-gated sodium channels enhance migration and invasiveness in cancer cells? Biochim. Biophys. Acta 1848, 2493–2501 (2015).
Roger, S., Gillet, L., Le Guennec, J. Y. & Besson, P. Voltage-gated sodium channels and cancer: is excitability their primary role? Front. Pharmacol. 6, 152 (2015).
Brackenbury, W. J. Voltage-gated sodium channels and metastatic disease. Channels (Austin) 6, 352–361 (2012).
Yang, M. et al. Therapeutic potential for phenytoin: targeting Nav1.5 sodium channels to reduce migration and invasion in metastatic breast cancer. Breast Cancer Res. Treat. 134, 603–615 (2012).
Fraser, S. P. et al. Voltage-gated sodium channel expression and potentiation of human breast cancer metastasis. Clin. Cancer Res. 11, 5381–5389 (2005).
Roger, S., Besson, P. & Le Guennec, J. Y. Involvement of a novel fast inward sodium current in the invasion capacity of a breast cancer cell line. Biochim. Biophys. Acta 1616, 107–111 (2003).
Driffort, V. et al. Ranolazine inhibits Nav1.5-mediated breast cancer cell invasiveness and lung colonization. Mol. Cancer 13, 264 (2014).
Nelson, M., Yang, M., Dowle, A. A., Thomas, J. R. & Brackenbury, W. J. The sodium channel-blocking antiepileptic drug phenytoin inhibits breast tumour growth and metastasis. Mol. Cancer 14, 13 (2015).
Gillet, L. et al. Voltage-gated sodium channel activity promotes cysteine cathepsin-dependent invasiveness and colony growth of human cancer cells. J. Biol. Chem. 284, 8680–8691 (2009).
Brisson, L. et al. NaV1.5 enhances breast cancer cell invasiveness by increasing NHE1-dependent H(+) efflux in caveolae. Oncogene 30, 2070–2076 (2011).
Brisson, L. et al. Nav1.5 Na+ channels allosterically regulate the NHE-1 exchanger and promote the activity of breast cancer cell invadopodia. J. Cell Sci. 126, 4835–4842 (2013).
Nelson, M., Millican-Slater, R., Forrest, L. C. & Brackenbury, W. J. The sodium channel beta1 subunit mediates outgrowth of neurite-like processes on breast cancer cells and promotes tumour growth and metastasis. Int. J. Cancer 135, 2338–2351 (2014).
Yu, F. H. et al. Sodium channel beta4, a new disulfide-linked auxiliary subunit with similarity to beta2. J. Neurosci. 23, 7577–7585 (2003).
Medeiros-Domingo, A. et al. SCN4B-encoded sodium channel beta4 subunit in congenital long-QT syndrome. Circulation 116, 134–142 (2007).
Tan, B. H. et al. Sudden infant death syndrome-associated mutations in the sodium channel beta subunits. Heart Rhythm 7, 771–778 (2010).
Okayama, H. et al. Identification of genes upregulated in ALK-positive and EGFR/KRAS/ALK-negative lung adenocarcinomas. Cancer Res. 72, 100–111 (2012).
Hou, J. et al. Gene expression-based classification of non-small cell lung carcinomas and survival prediction. PLoS ONE 5, e10312 (2010).
Chioni, A. M., Brackenbury, W. J., Calhoun, J. D., Isom, L. L. & Djamgoz, M. B. A novel adhesion molecule in human breast cancer cells: voltage-gated Na+ channel beta1 subunit. Int. J. Biochem. Cell Biol. 41, 1216–1227 (2009).
Marques, I. J. et al. Metastatic behaviour of primary human tumours in a zebrafish xenotransplantation model. BMC Cancer 9, 128 (2009).
Jelassi, B. et al. P2X(7) receptor activation enhances SK3 channels- and cystein cathepsin-dependent cancer cells invasiveness. Oncogene 30, 2108–2122 (2011).
Roger, S., Guennec, J. Y. & Besson, P. Particular sensitivity to calcium channel blockers of the fast inward voltage-dependent sodium current involved in the invasive properties of a metastastic breast cancer cell line. Br. J. Pharmacol. 141, 610–615 (2004).
Wannous, R. et al. Suppression of PPARbeta, and DHA treatment, inhibit NaV1.5 and NHE-1 pro-invasive activities. Pflugers Arch. 467, 1249–1259 (2015).
Jarvis, M. F. et al. A-803467, a potent and selective Nav1.8 sodium channel blocker, attenuates neuropathic and inflammatory pain in the rat. Proc. Natl. Acad. Sci. USA 104, 8520–8525 (2007).
Roger, S. et al. Voltage-gated sodium channels potentiate the invasive capacities of human non-small-cell lung cancer cell lines. Int. J. Biochem. Cell Biol. 39, 774–786 (2007).
Diss, J. K., Archer, S. N., Hirano, J., Fraser, S. P. & Djamgoz, M. B. Expression profiles of voltage-gated Na+ channel alpha-subunit genes in rat and human prostate cancer cell lines. Prostate 48, 165–178 (2001).
Saltel, F. et al. Invadosomes: intriguing structures with promise. Eur. J. Cell Biol. 90, 100–107 (2011).
Kovacs, M., Toth, J., Hetenyi, C., Malnasi-Csizmadia, A. & Sellers, J. R. Mechanism of blebbistatin inhibition of myosin II. J. Biol. Chem. 279, 35557–35563 (2004).
Messner, D. J. & Catterall, W. A. The sodium channel from rat brain. Separation and characterization of subunits. J. Biol. Chem. 260, 10597–10604 (1985).
Qin, N. et al. Molecular cloning and functional expression of the human sodium channel beta1B subunit, a novel splicing variant of the beta1 subunit. Eur. J. Biochem. 270, 4762–4770 (2003).
Brackenbury, W. J. & Isom, L. L. Na channel beta subunits: overachievers of the ion channel family. Front. Pharmacol. 2, 53 (2011).
McCormick, K. A. et al. Molecular determinants of Na+ channel function in the extracellular domain of the beta1 subunit. J. Biol. Chem. 273, 3954–3962 (1998).
Meadows, L., Malhotra, J. D., Stetzer, A., Isom, L. L. & Ragsdale, D. S. The intracellular segment of the sodium channel beta1 subunit is required for its efficient association with the channel alpha subunit. J. Neurochem. 76, 1871–1878 (2001).
Chen, C. et al. Identification of the cysteine residue responsible for disulfide linkage of Na+ channel alpha and beta2 subunits. J. Biol. Chem. 287, 39061–39069 (2012).
Gilchrist, J., Das, S., Van Petegem, F. & Bosmans, F. Crystallographic insights into sodium-channel modulation by the beta4 subunit. Proc. Natl. Acad. Sci. USA 110, E5016–E5024 (2013).
Calhoun, J. D. & Isom, L. L. The role of non-pore-forming beta subunits in physiology and pathophysiology of voltage-gated sodium channels. Handb. Exp. Pharmacol. 221, 51–89 (2014).
Lenkowski, P. W., Shah, B. S., Dinn, A. E., Lee, K. & Patel, M. K. Lidocaine block of neonatal Nav1.3 is differentially modulated by co-expression of beta1 and beta3 subunits. Eur. J. Pharmacol. 467, 23–30 (2003).
Zhang, M. M. et al. Co-expression of Navβ subunits alters the kinetics of inhibition of voltage-gated sodium channels by pore-blocking mu-conotoxins. Br. J. Pharmacol. 168, 1597–1610 (2013).
Wilson, M. J. et al. Navβ subunits modulate the inhibition of Nav1.8 by the analgesic gating modifier muO-conotoxin MrVIB. J. Pharmacol. Exp. Ther. 338, 687–693 (2011).
Isom, L. L. Sodium channel beta subunits: anything but auxiliary. Neuroscientist 7, 42–54 (2001).
Isom, L. L. & Catterall, W. A. Na+ channel subunits and Ig domains. Nature 383, 307–308 (1996).
Isom, L. L. The role of sodium channels in cell adhesion. Front. Biosci. 7, 12–23 (2002).
Hernandez-Plata, E. et al. Overexpression of Nav1.6 channels is associated with the invasion capacity of human cervical cancer. Int. J. Cancer. 130, 2013–2023 (2012).
Jezequel, P. et al. bc-GenExMiner: an easy-to-use online platform for gene prognostic analyses in breast cancer. Breast Cancer Res. Treat. 131, 765–775 (2012).
Gore, J., Besson, P., Hoinard, C. & Bougnoux, P. Na(+)-H+ antiporter activity in relation to membrane fatty acid composition and cell proliferation. Am. J. Physiol. 266, C110–C120 (1994).
Soderberg, O. et al. Direct observation of individual endogenous protein complexes in situ by proximity ligation. Nat. Methods 3, 995–1000 (2006).
Westerfield, M. The Zebrafish Book: A Guide for the Laboratory Use of Zebrafish (Danio rerio) 5th edn (Univ. Oregon Press, 2007)...
This work was supported by the ‘Ministère de la Recherche et des Technologies’, the Inserm, the ‘Ligue Nationale Contre le Cancer–Interrégion Grand-Ouest’, the Région Centre (grant ‘NaVMetarget’, project ‘ARD2020 Biomédicaments’) and the ‘Association CANCEN’. J.L.G.-P. was supported by ‘Le Studium’. We thank Mrs Catherine Le Roy for secretary and administrative assistance, and Dr Armelle Vinceneux for her help with IHC staining of β4 in human breast normal tissues. We thank Prof. Stephan J. Reshkin (University of Bari, Italy), Dr Lin-Hua Jiang (University of Leeds, UK) and Dr Philippe Chavrier (Institut Curie, France) for their critical reading of the manuscript.
The authors declare no competing financial interests.
About this article
Cell Calcium (2019)
Phosphorylation of NHERF1 S279 and S301 differentially regulates breast cancer cell phenotype and metastatic organotropism
Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease (2019)
Contribution of voltage-gated sodium channel β-subunits to cervical cancer cells metastatic behavior
Cancer Cell International (2019)
Frontiers in Oncology (2019)
Genetic cooperativity in multi-layer networks implicates cell survival and senescence in the striatum of Huntington’s disease mice synchronous to symptoms