Functional expression of the voltage-gated Na+-channel Nav1.7 is necessary for EGF-mediated invasion in human non-small cell lung cancer cells

Summary Various ion channels are expressed in human cancers where they are intimately involved in proliferation, angiogenesis, invasion and metastasis. Expression of functional voltage-gated Na+ channels (Nav) is implicated in the metastatic potential of breast, prostate, lung and colon cancer cells. However, the cellular mechanisms that regulate Nav expression in cancer remain largely unknown. Growth factors are attractive candidates; they not only play crucial roles in cancer progression but are also key regulators of ion channel expression and activity in non-cancerous cells. Here, we examine the role of epidermal growth factor receptor (EGFR) signalling and Nav in non-small cell lung carcinoma (NSCLC) cell lines. We show unequivocally, that functional expression of the &agr; subunit Nav1.7 promotes invasion in H460 NSCLC cells. Inhibition of Nav1.7 activity (using tetrodotoxin) or expression (by using small interfering RNA), reduces H460 cell invasion by up to 50%. Crucially, non-invasive wild type A549 cells lack functional Nav, whereas exogenous overexpression of the Nav1.7 &agr; subunit is sufficient to promote TTX-sensitive invasion of these cells. EGF/EGFR signalling enhances proliferation, migration and invasion of H460 cells but we find that, specifically, EGFR-mediated upregulation of Nav1.7 is necessary for invasive behaviour in these cells. Examination of Nav1.7 expression at mRNA, protein and functional levels further reveals that EGF/EGFR signalling via the ERK1/2 pathway controls transcriptional regulation of channel expression to promote cellular invasion. Immunohistochemistry of patient biopsies confirms the clinical relevance of Nav1.7 expression in NSCLC. Thus, Nav1.7 has significant potential as a new target for therapeutic intervention and/or as a diagnostic or prognostic marker in NSCLC.


Introduction
Lung cancer is the most prevalent cancer world-wide. Small cell lung carcinoma (SCLC) and non-small cell lung carcinoma (NSCLC) occur with a frequency of ,20% and ,80%, respectively. NSCLC has a low, average 5-year, survival rate (14%) and the development of metastases results in extremely poor prognosis (Jemal et al., 2011). The epidermal growth factor receptor (EGFR) plays key roles in the progression of various cancers (Araújo et al., 2007;Cohen, 2002;Mendelsohn, 2003), and in NSCLC its overexpression in ,40-80% of patients is also associated with poor prognosis (Mendelsohn, 2003). Smallmolecule EGFR kinase inhibitors (gefitinib and erlotinib) have recently shown significant therapeutic benefit in many patients with advanced NSCLC, particularly those with somatic mutations of the EGFR (Araújo et al., 2007). However, drug resistance, whether inherent or acquired, remains a main cause of chemotherapy failure (Cataldo et al., 2011;Domingo et al., 2010;Lynch et al., 2004;Paez et al., 2004). Consequently, there is a pressing need for improved treatments based upon improved understanding of the effector pathways downstream of EGFR signalling.
In addition to their roles in cell growth and cancer progression, growth factors also regulate the expression and activity of numerous ion channels (Akopian et al., 1999; This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed. Goetz et al., 2009;Goldfarb et al., 2007;Lei et al., 2001;Lou et al., 2005;Toledo-Aral et al., 1995). EGF enhances Na v currents in non-cancerous cells, including cardiac muscle (Liu et al., 2007). In human and rat prostate cancer, functional upregulation of Na v is implicated in the pro-invasive response of cells to EGF, although the cellular mechanism(s) involved are unknown (Ding et al., 2008;Uysal-Onganer and Djamgoz, 2007). In the present study, we conducted a detailed characterisation of Na v in NSCLC and examined the effects of EGF/EGFR signalling on expression and activity of Na v . Briefly, we show that functional expression of the Na v 1.7 a subunit isoform specifically promotes invasion of strongly invasive H460 NSCLC cells. Moreover, non-invasive A549 NSCLC cells that lack Na v , can be induced to invade simply by overexpression of Na v 1.7. In strongly invasive H460 cells, EGF/EGFR signalling through the ERK1/2 pathway controls transcriptional regulation of Na v 1.7 to promote invasion. Finally, immunohistochemistry of patient biopsies provides evidence that supports the clinical relevance of Na v 1.7 expression.

Results
To assess the role of Na v in NSCLC it was first necessary to identify suitable model cell lines for comparison of Na v activity and expression in weakly versus strongly invasive cells. We, therefore, tested 22 NSCLC cell lines for functional Na v expression using a high-throughput IonWorksH Quattro TM Automated Patch Clamp System to record the inward Na v current (I Na ) (Fig. 1A). Two cell lines were selected for further study: A549 cells that lacked I Na and H460 cells that exhibited the most-robust I Na both in terms of current amplitude and proportion of cells with current (,45%, n589). Conventional patch-clamp recordings confirmed the IonWorksH data ( Fig. 1B,C). H460 cells were more proliferative (Fig. 1D) and more invasive (Fig. 1F) than A549 cells. However, migration was greater in A549 cells (Fig. 1E). H460 cells were, thus, chosen as a strongly invasive model, whereas A549 cells were used as a model of weak invasive potential. These preliminary findings agree with those described by Roger et al. (Roger et al., 2007).

Na v promotes H460 cell invasion
To explore the link between functional expression of Na v and invasion of H460 cells, we used the Na v -specific inhibitor (tetrodotoxin, TTX) or activator (veratridine) that, respectively, block or enhance the opening of Na v channels. TTX (0.5 mM) abolished I Na , whereas 50 mM veratridine induced a modest increase in the persistent inward Na + current (I Na,P ) ( Fig. 2A). Neither TTX nor veratridine affected proliferation or migration (Fig. 2B,C). However, as predicted, blocking Na v channels (using TTX) reduced invasion of H460 cells by 39.767.2%, whereas enhanced opening of Na v (using veratridine) promoted invasion (19.764.7%) (Fig. 2D). Thus, Na v activity is specifically associated with invasion in these cells. We next generated two H460 dilution clones on the basis of high versus low Na v activity (Fig. 2E). In agreement with the pharmacology, invasion was enhanced in the 'high' compared with the 'low' clone cells (Fig. 2D). The relatively modest enhancement of invasion seen in WT H460 cells treated with veratridine, or, in the high versus low clone cells, reflects the similarly modest increases in current density.
Na v are usually expressed in excitable cells (neurons and muscle), where they are activated by depolarisation of the cell membrane in response to action potentials (Catterall, 2000). However, cancerous cells are most commonly of epithelial origin (non-excitable), raising the question; how are Na v activated? In other invasive cancer cells, a small I Na window current at the resting membrane potential (E m ) of cells has been postulated to produce sufficient Na + influx to promote invasion (Gillet et al.,Fig. 1. Characterisation of H460 and A549 NSCLC cell lines. (A) Histogram showing average maximum peak current amplitude I max for inward Na + current and uncharacterised outward currents in NSCLC cell lines at 210, 0 and +10 mV. Currents were recorded using the IonWorksH Quattro TM Automated Patch Clamp System (n$40 for all cell lines). (B) Average current-density-voltage (I-V) curves for H460 (n517) and A549 (n516) cells. Continuous lines indicate Boltzmann curve fits using the Boltzmann function described in Materials and Methods. Currents were evoked using 30-millisecond depolarising voltage steps in 5-mV intervals (290 to +70 mV) from a holding potential V h 5 290 mV. (C) Representative current traces from H460 and A549 cells. (D) Proliferation of H460 and A549 cells, recorded over 144 hours, measured as percent change in confluence from time 0 hours (n516 wells per cell line and at least two separate experiments). (E) H460 and A549 cell migration, recorded over 58 hours using the scratch-wound assay, measured as percent change in wound confluence (n58 wells per cell line, two separate experiments). (F) Matrigel TM invasion of H460 and A549 cells, recorded after 48 hours (n59 for both cell lines, ***P,0.001, Student's t-test).
2009; Roger et al., 2007). As shown in Fig. 2F, H460 cells may also possess a small I Na window current (between 240 and 210 mV), where Na v are partially activated but not fully inactivated. Notably, the E m of these cells (22764 mV) lies within this voltage range, thus, allowing Na + influx (Fig. 2G). However, the more hyperpolarised E m of weakly invasive A549 cells (23663 mV) is consistent with the lack of Na v channel activity in these cells. TTX-induced Na v block hyperpolarised the E m of H460 cells to 23761 mV (similar to that of A549 cells), where Na + influx is predicted to be negligible. By contrast, when the cell membrane was depolarised to 22361 mV by using veratridine, I Na,P was enhanced. Consistent with these findings, the intracellular Na + concentration [Na + ] i , was ,2-fold higher in H460 than A549 cells (22.364.3 mM versus 9.763.2 mM; Fig. 2H). Moreover, in TTX-treated H460 cells, [Na + ] i was reduced to 10.862.8 mM, although veratridine had no significant effect (28.864.0 mM).
Na v 1.7 underlies Na v current and promotes the invasive capability of H460 cells Na v channels typically comprise one pore-forming a subunit and one or two auxiliary b subunits that modulate channel gating and cell surface expression (Catterall, 2000;Joho et al., 1990). b subunits, particularly b1, also function as cell adhesion molecules (CAMs) (Brackenbury and Isom, 2008;Chioni et al., 2009;Isom, 2001;Isom, 2002;Patino and Isom, 2010). Nine aand four bsubunit isoforms have been identified (Catterall et al., 2005). Of these, Na v 1.5, Na v 1.8 and Na v 1.9 are TTX-resistant (EC 50 55.7-60 mM), whereas the remainder are TTX-sensitive (EC 50 54-25 nM). Since 0.5 mM TTX abolished I Na in H460 cells, we measured mRNA levels for all TTX-sensitive Na v a subunits, using quantitative PCR (qPCR) (Fig. 3A). We also assessed mRNA expression of the four known b subunits (Fig. 3A). In strongly invasive H460 cells, Na v 1.7 mRNA expression was ,4-fold higher compared with other a subunit isoforms, at levels comparable with functional expression of Na v in SH-SY5Y neuroblastoma cells. Low mRNA expression of all Na v a subunits was seen in A549 cells, consistent with their lack of I Na . Western immunoblotting (Fig. 3B) and immunocytochemistry (Fig. 3C) also confirmed the presence of Na v 1.7 protein in H460 but not in A549 cells. Furthermore, Na v 1.7 mRNA expression was ,60% lower in the 'low' versus 'high' H460 dilution clone, consistent with the difference in I Na (Fig. 3D). Transfection of H460 cells with small interfering RNA (siRNA) against Na v 1.7 (SCN9A) completely abolished I Na (Fig. 3E,F) and reduced the invasive capability of the cells by 50.261.8% (Fig. 3G), indicating that the functional expression of Na v 1.7 is necessary for Na v -mediated invasion of H460 cells.
H460 cells exhibited very low expression of all Na v b subunit mRNAs (Fig. 3A), whereas A549 cells expressed high levels of b1 mRNA (,8-fold more than H460 cells; Fig. 3A). Consistent with a role for the Na v b subunit in cell adhesion (Brackenbury and Isom, 2008;Chioni et al., 2009;Isom, 2001;Isom, 2002;Patino and Isom, 2010), in A549 cells, adhesion was invasion, after 48 hours. Invasion in two H460 dilution clones with high versus low Na v expression and activity are shown for comparison. n59 for all conditions; *P,0.05, **P,0.01, one-way ANOVA and SNK correction. (E) Average current-density-voltage (I-V) relationships for Na v currents in high (open square, n59) and low (closed square, n510) H460-cell dilution clones, recorded using conventional patch clamp technique (*P,0.05, Student's t-test). Continuous lines indicate Boltzmann curve fits using as described in Materials and Methods. Inset: representative current traces from high versus low clone cells. Currents were evoked using 30-millisecond depolarising voltage steps in 5-mV intervals (290 to +70 mV), from V h 5 290 mV. (F) Average conductance-voltage (conductance; squares, n517) and steady-state inactivation-voltage (availability; circles, n510) curves for I Na in H460 cells. Continuous lines indicate Boltzmann curve fits using the functions described in Materials and Methods. (G) Average resting membrane potential E m in H460 cells, A549 cells and H460 cells treated with 0.5 mM TTX or 50 mM veratridine (Ver) for 2-3 minutes. n510 for all conditions (*P,0.05, **P,0.01, ***P,0.001, one-way ANOVA and SNK correction). (H) Intracellular Na + concentration [Na + ] i , in control versus TTX-treated (0.5 mM) or veratridinetreated (50 mM, Ver) H460 cells and control A549 cells. [Na + ] i was monitored using the ratiometric fluorescent dye SBFI (n55 for all conditions, **P,0.01, one-way ANOVA and SNK correction). CTL, untreated control H460 cells.
approximately double than that of H460 cells (Fig. 4B). Moreover, inhibiting the expression of the b1 subunit by using siRNA against SCN1B resulted in a modest decrease in adhesion and a concurrent increase in invasion compared with wild-type A549 cells (Fig. 4B,C). Notably, the invasive capability of H460 cells was reduced simply by overexpressing b1-GFP (Fig. 4D,F). Conversely, in A549 cells, TTX-sensitive invasion was enhanced by overexpression of Na v 1.7-DsRed alone (Fig. 4E,F). Thus, functional expression of Na v 1.7 provides a driving force for NSCLC cell invasion, whereas the b1 subunit promotes cell adhesion and reduced invasion.
Na v 1.7 expression is enhanced in cancerous lung tissue The clinical relevance of our findings was assessed by comparing Na v 1.7 protein expression in cancerous versus normal-matched lung tissue from patient biopsies. Na v 1.7 immunostaining was markedly higher in tumour versus normal lung tissue for all patient samples tested ( Fig. 5A-C), consistent with Na v 1.7 expression in H460 versus A549 cells (Fig. 5D). Notably, Fig. 5C clearly shows disorganised tumour tissue bordering healthy lung tissue, with a marked difference in the levels of immunostaining for Na v 1.7. Increased Na v 1.7 protein expression may thus accompany the transition from low-to high-grade tumours. Differently staged lung cancer samples could not be obtained, but comparison of prostate cancer tumour samples showed negligible immunostaining for Na v 1.7 in low-grade tumours, whereas in high-grade tumours Na v 1.7 protein levels were high (Fig. 5E). Like NSCLC, metastatic prostate cancer is also associated with functional expression of Na v 1.7 (Diss et al., 2001;Diss et al., 2005;Nakajima et al., 2009;Yildirim et al., 2012). Together, these data support a role for Na v 1.7 in the development of an invasive cancer phenotype in vivo.
EGF upregulates Na v 1.7 expression and H460 cell invasion EGF/EGFR signalling plays key roles in cancer progression (Araújo et al., 2007;Cohen, 2002;Mendelsohn, 2003). In Fig. 3. Na v 1.7 is necessary for Na v channel-mediated invasion in H460 cells. (A) Relative mRNA expression for all TTX-sensitive Na v a subunits and all auxiliary Na v b subunits in A549 and H460 NSCLC cells, and in positive control SH-SY5Y cells. All expression data were normalised to 18S rRNA expression (n56 for all primer sets, *P,0.05, Student's t-test). (B) Representative western immunoblots showing expression of Na v 1.7 (upper) and b-actin (lower) proteins in Na v 1.7-negative (HEK-293) and Na v 1.7-positive (HEK-293 cells stably transfected with Na v 1.7, HEK-Na v 1.7) control cells, A549 cells, H460 cells and H460 cells transfected with 5 nM siRNA directed against Na v 1.7 (n53 for all blots). (C) Representative fluorescence images showing cellular distribution of Na v 1.7 protein in Na v 1.7-positive (HEK-Na v 1.7) and Na v 1.7-negative (HEK-293) control cells, and in H460 versus A549 cells. Na v 1.7 was stained using a Na v 1.7-specific antibody recognising a C-terminal intracellular epitope (green). CTL 1, no primary antibody incubation; CTL 2, no secondary antibody incubation (n$10 for all cell lines; scale bars: 10 mm; blue, DAPI nuclear staining. (D) Relative Na v 1.7 mRNA expression in high and low H460 cell dilution clones (n56 for both clones, **P ,0.01, Student's t-test). Inset: representative western immunoblots showing Na v 1.7 (upper) and b-actin (lower) protein expression in high and low dilution clones (n53 for both blots). (E) Relative Na v 1.7 mRNA expression for control H460 cells (CTL), H460 cells transfected with 5 nM scrambled siRNA (Scrambled) and H460 cells transfected with 5 nM siNa v 1.7 (n56 for all conditions, ***P,0.001, one-way ANOVA and SNK correction).
NSCLC, EGFR overexpression is associated with particularly aggressive malignancies (Mendelsohn, 2003). To assess the role of EGFR signalling in the upregulation of Na v 1.7 and the subsequent invasion in NSCLC, we initially compared Na v 1.7 mRNA expression in H460 cells following treatments with EGF or EGFR signalling modulators (Fig. 6A). Serum starvation (48 hours) reduced Na v 1.7 mRNA levels by 81.261.1% compared with control cells maintained in complete medium.
In contrast, addition of EGF (100 ng.ml 21 , 24 hours) following 24 hour serum starvation, fully restored Na v 1.7 mRNA expression. Conversely, the EGFR inhibitor gefitinib (1 mM, 24 hours), in complete medium reduced Na v 1.7 mRNA expression to serum-starved levels. Moreover, gefitinib blocked EGF-mediated enhancement of Na v 1.7 mRNA expression in these cells. The Na v 1.7 protein levels mirrored these changes in mRNA expression (inset Fig. 6A).
To avoid adverse effects of serum starvation on the integrity of cell membranes, electrophysiology was performed on H460 cells maintained in the presence of serum. Under these conditions, EGF (24 hours) had a negligible effect on I Na (Fig. 6B), presumably because EGF-mediated signalling is already close to its maximum under control conditions. However, in the presence of either gefitinib or anti-EGF function-blocking antibody (10 mg.ml 21 ), I Na was almost completely abolished. The concurrent changes in Na v 1.7 mRNA, protein and I Na (current density, pA.pF 21 ) suggest that EGF/EGFR signalling controls transcriptional regulation of Na v 1.7 and, hence, numbers of functional channels at the cell surface. Growth factor signalling can, however, induce rapid post-translational regulation of ion channels within minutes, e.g. phosphorylation of channels at the cell surface (Martin et al., 2006;Woodall et al., 2008) and/or rapid trafficking of channels to the cell surface (Kanzaki et al., 1999). However, because acute (10-minute) application of gefitinib had no effect on I Na (Fig. 6C), we conclude that endogenous EGF/EGFR signalling controls transcriptional rather than post-translational regulation of Na v 1.7. Fig. 4. Reciprocal expression of Na v 1.7 and b1 subunits regulates invasion and adhesion. (A) Relative expression of Na v b1 subunit mRNA in control A549 cells versus A549 cells transfected with 5 nM siRNA against Na v b1 (siSCN1B), or scrambled siRNA (n56 for all conditions, ***P,0.001, one-way ANOVA and SNK correction). (B) Effect of b1-siRNA on A549 cell adhesion, analysed by Crystal Violet staining (n58 for all conditions, **P,0.01, ***P,0.001, one-way ANOVA and SNK correction). Adhesion in H460 cells is shown for comparison. (C) Effect of siSCN1B on A549 cell Matrigel TM invasion after 48 hours (n59 for both conditions, *P,0.05, Student's t-test).
(D) Relative invasion into Matrigel TM of H460 cells transfected with either control GFP (CTL) or b1-GFP (H460 + b1-GFP) (n54 for all conditions, **P ,0.01, Student's t-test). (E) Relative invasion into Matrigel TM of A549 cells transfected with GFP control (CTL) or Na v 1.7-DsRed in the absence (A549 + Na v 1.7-DsRed) or presence of 0.5 mM TTX (A549 + Na v 1.7-DsRed + TTX). n54 for all conditions, **P ,0.01, one-way ANOVA and SNK correction). (F) Fluorescence images for H460 cells (left) and A549 cells (right) transfected with b1-GFP and Na v 1.7-DsRed, respectively (scale bars: 50 mm). H460 cell proliferation, migration and invasion were all upregulated by EGF/EGFR signalling, whereas Na v contributed only to invasion (Fig. 6D-F). Since inhibition of Na v (TTX), EGFR (gefitinib), or combined EGF/TTX treatments all result in ,40% inhibition of invasion compared with control untreated cells, we infer that the pro-invasive response to EGF is mediated exclusively via Na v activity. Notably, however, neither gefitinib nor TTX completely suppressed invasion. Thus, although EGF/EGFR/Na v signalling accounts for ,40-50% of invasion, other cellular factors must also contribute to the pro-invasive capabilities of these cells.
Finally, we examined the involvement of the signalling pathways downstream of EGFR. Four main pathways are associated with EGFR signalling: Jak/Stat, PLCc, PI3-K/Akt and ERK1/2 (Oda et al., 2005). Of these, we focused on the PI3-K/Akt and ERK1/2 signalling cascades, which modulate ion channels in excitable cells (Adams et al., 2000;Black et al., 2008;Persson et al., 2011;Qiu et al., 2003;Smani et al., 2010;Stamboulian et al., 2010;Woodall et al., 2008) and are intimately involved in cancer progression (Frémin and Meloche, 2010;Keshet and Seger, 2010;Li et al., 2005;Stahl et al., 2004). Western immunoblotting confirmed basal activity of both phospho-ERK1/2 and phospho-Akt in H460 cells (Fig. 7A). U0126 inhibited Na v 1.7 mRNA expression but wortmannin had no effect, implying that ERK1/2 (not PI3-K) controls Na v 1.7 expression (Fig. 7B). Effects on I Na were initially screened using IonWorksH Automated Patch Clamp (Fig. 7C). As with Na v 1.7 mRNA expression, U0126 inhibited mean I Na by 44% from 0.3960.05 nA (control) to 0.2260.02 nA, whereas wortmannin application had no effect. These data were further supported by conventional patch clamp recordings (Fig. 7D). Consistent with these effects on Na v 1.7 expression and activity, H460 cell invasion was unaffected by wortmannin but inhibited by U0126 (Fig. 7E). No additive effect of combining U0126 with TTX was observed, suggesting that regulation via ERK1/2 is crucial for Na v 1.7-mediated invasion of H460 cells.

Discussion
We have shown that EGF/EGFR signalling via a U0126-sensitive ERK1/2 pathway controls transcriptional upregulation of Na v 1.7 to promote cellular invasion in NSCLC cell lines. Moreover, immunohistochemistry of patient biopsies provides evidence supporting the clinical relevance of Na v 1.7 protein expression in NSCLC. Thus, we identify Na v 1.7 as a new downstream effector of the EGF/EGFR pathway with potential as a target for therapeutic intervention and/or as a diagnostic or prognostic marker in NSCLC.
The necessity of Na v 1.7 functional expression for invasion of NSCLC cells agrees with other work showing the crucial role of Na v -particularly Na v 1.5 and Na v 1.7 -in metastatic behaviours of several cancers, including NSCLC. Na v 1.7 mRNA expression in H460 cells has been reported (Roger et al., 2007) but its predominant role in Na v -dependent invasion of these cells has not been shown before. In breast cancer, the Na v b1 subunit is reported to act as a CAM, being highly expressed in non-invasive cells but not in strongly invasive cells that exhibit reduced cell adhesion (Brackenbury and Isom, 2008;Chioni et al., 2009). We see similar trends in the reciprocal expression of the Na v b1 and Na v 1.7 subunits in non-invasive A549 versus highly invasive H460 NSCLC cells. Moreover, regulation of Na v b1 and Na v 1.7 expression alone is sufficient to alter significantly the invasive properties of H460 and A549 cells. Thus, the balance of Na v b1 versus a subunit expression also appears to regulate the invasive potential of NSCLC cells.
To date, there has been little detailed characterisation of the mechanisms that drive functional expression of Na v in cancerous cells (Ding et al., 2008;Fraser et al., 2010;Uysal-Onganer and Fig. 7. EGF upregulates Na v 1.7 expression and the subsequent H460 cell invasion through ERK1/2 signalling. (A) Representative western immunoblots showing the levels of active phospho-ERK1/2 (pERK1/2) versus total ERK (top), and active phospho-Akt (pAkt) versus total Akt (bottom) proteins in control untreated (CTL), U0126-(10 mM) and wortmannin-treated (100 nM, Wort) H460 cells after 24 hours (n53 for all blots). (B) Relative Na v 1.7 mRNA expression in H460 cells showing the effect of treatment with U0126 (10 mM) or wortmannin (100 nM, Wort) for 24 hours compared with control untreated (CTL) H460 cells (n56 for all conditions, *P,0.05, one-way ANOVA and SNK correction). (C) Maximum current amplitudes (I max ) from individual H460 cells that had been left untreated (CTL), or had been treated with U0126 (10 mM) or wortmannin (100 nM, Wort), were recorded using the IonWorksH Quattro TM Automated Patch Clamp System (n5number of cells with TTX-inhibitable I Na /total number of cells tested). (D) Average current-density-voltage (I-V) plots for control untreated (open squares, n59) and 10 mM U0126-treated (closed squares, n511) H460 cells (**P,0.01, Student's t-test). Continuous lines indicate Boltzmann curve fits as described in Materials and Methods. Inset: representative current traces from untreated (CTL) and U0126-treated H460 cells. Currents were evoked using 30-millisecond depolarising steps in 5-mV intervals (290 to +70 mV), from V h 5290 mV. (E) Relative Matrigel TM invasion of H460 cells after 48 hours in control untreated conditions (CTL) and following treatment with 10 mM U0126, 10 mM U0126/1 mM TTX, 100 nM wortmannin (Wort) and 100 nM wortmannin/1 mM TTX (Wort + TTX) (n59 for all conditions, *P,0.05, **P,0.01, one-way ANOVA and SNK correction). Na v 1.6 GCGTTGAGGCACTACTACTTC TCCCACAATGGAGAGGATGAC SCN9A Na v 1.7 AGACCTCTCTTTCCATGTAGATTAC TGTAACTGCCTTTCTGTATTGTTG , 2007). To address this crucial issue, we focused on the role of EGF/EGFR signalling, which not only plays key roles in NSCLC (Fujimoto et al., 2005;Mendelsohn, 2003;Volante et al., 2007) but also regulates transcription and activity of various ion channels, including Na v (Adams et al., 2000;Black et al., 2008;Ding et al., 2008;Liu et al., 2007;Persson et al., 2011;Qiu et al., 2003;Smani et al., 2010;Stamboulian et al., 2010;Uysal-Onganer and Djamgoz, 2007;Woodall et al., 2008). In H460 cells, our data suggest that EGF/EGFR signalling via ERK1/2 tonically upregulates the functional expression of Na v 1.7 to enhance Na + influx and, hence, invasion. Whereas proliferation, migration and invasion of H460 cells are all promoted via the EGF/EGFR pathway, only invasion is sensitive to Na v 1.7 inhibition. Since blocking of EGFR, ERK1/2 or Na v 1.7 inhibits invasion by ,40-50% -with no additive effects of combined treatments -we infer that EGF/EGFR-ERK-mediated invasion in these cells is most likely to occur via upregulation of Na v 1.7. However, since this pathway promotes only 40-50% of H460 cell invasion, other cellular factors must also be involved. Vascular endothelial, fibroblast and platelet-derived growth factors all play roles in NSCLC (Ballas and Chachoua, 2011) and are linked with Na v function (Andrikopoulos et al., 2011;Goldfarb et al., 2007;Hilborn et al., 1998). Although further work is required to establish the relevance of such growth factors, it is conceivable that Na v 1.7 is a convergence point in several cancer signalling pathways. Whereas transcriptional upregulation is necessary to promote sufficient functional expression of Na v protein at the cell surface, our data suggest it is the influx of Na + through Na v 1.7 and the consequent depolarisation of the cell membrane, which appears to be essential for invasion. Although inhibition of EGF/EGFR, ERK1/2 and Na v 1.7 consistently reduced invasion by ,40-50%, the effects on I Na varied from complete loss of current (TTX) to 47% inhibition (U0126). This implies that not all available channels need to be functional to enable sufficient Na + influx for invasion. Indeed, in agreement with others (Gillet et al., 2009;Roger et al., 2007), we also found evidence of continuous Na + influx in H460 cells, when Na v channels are partially activated at resting membrane potential. Furthermore, block of Na v activity hyperpolarises the plasma membrane and reduces [Na + ] i , whereas enhanced channel opening depolarises the membrane potential and raises [Na + ] i . Thus, a positive-feedback mechanism appears to operate, where the partial opening of Na v at resting potential allows Na + influx, which -in turn -keeps the plasma membrane sufficiently depolarised to maintain a continuous influx of Na + . Under physiological conditions, this influx of Na + is sufficient to maintain the invasive capability of cells. Endogenous modulators of Na v 1.7, e.g. ERK1/2, can thus provide an exquisite mechanism for fine-tuning of EGF/EGFR-mediated invasion in NSCLC. Similarly, other transport proteins that allow Na + influx are also expected to promote invasive behaviour in NSCLC. For example, in prostate cancer cells, overexpression of exogenous Na v 1.4 enhances their invasion (Bennett et al., 2004). In MDA-MB-231 breast cancer cells, Na v 1.5 is postulated to be functionally coupled with the Na + /H + exchanger NHE1. This exchanger extrudes H + to cause extracellular acidification, thereby providing an optimal environment for pH-dependent activity of cysteine cathepsins and proteolysis of the ECM, leading to enhanced cell invasion (Brisson et al., 2011;Gillet et al., 2009). Na + influx has also been suggested to impact on intracellular Ca 2+ homeostasis, primarily through reversal of Na + /Ca 2+ exchanger NCX activity (Blaustein and Lederer, 1999). This could have numerous effects, including altered gene expression.
Regardless of how increased [Na + ] i promotes invasion, we have demonstrated unequivocally that functional expression of Na v 1.7 is associated with increased invasive potential of NSCLC cell lines. Whereas the use of small-molecule EGFR inhibitors, such as gefitinib and erlotinib, has substantial therapeutic benefits for many patients with advanced NSCLC, drug resistance remains a significant problem (Cataldo et al., 2011;Domingo et al., 2010;Lynch et al., 2004;Paez et al., 2004). Drugs that target Na v 1.7 activity could, thus, prove useful in conjunction with existing therapies. Na v -specific blockers are already in clinical use as local anaesthetics and, more recently, Na v 1.7specific inhibitors are being developed for the treatment of pain (Bregman et al., 2011;Chowdhury et al., 2011;Clare, 2010;Ghelardini et al., 2010). Given its functional significance and marked upregulation in tumour tissue, we suggest that Na v 1.7 represents an important new target for therapeutic intervention and/or as a diagnostic and/or prognostic marker in NSCLC.
Cell culture A549 and H460 cells were obtained from AstraZeneca, UK, were genetically tested and authenticated using the PowerPlexH 16 System (Promega, UK) and were not cultured for more than 6 months. A549 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% foetal bovine serum (FBS; Gibco/Invitrogen, UK) and 50 U.ml 21 penicillin/50 mg.ml 21 streptomycin (Sigma-Aldrich, UK). H460 cells were cultured in RPMI-1640 medium (Gibco/ Invitrogen, UK) supplemented with FBS and penicillin/streptomycin. Cells were maintained at 37˚C, 5% CO 2 . Cells were seeded and maintained in complete medium at least overnight, prior to any treatment. All pharmacological agents were dissolved in appropriate solvents and diluted in cell culture medium.

Electrophysiology
Recordings of inward Na v channel current, I Na , from A549 and H460 cells were made using the whole-cell patch clamp technique. The extracellular solution contained (in mM): NaCl 140, CsCl 3, MgCl 2 1, CaCl 2 2, glucose 11, HEPES 10 adjusted to pH 7.4 with NaOH and 320 mOsm.litre 21 with sucrose. The internal solution contained (in mM): CsCl 135, NaCl 5, CaCl 2 0.37, MgCl 2 1, HEPES 10, EGTA 5, adjusted to pH 7.2 with CsOH and 310 mOsm.litre 21 with sucrose. Before recording, cells were replated onto 0.01% poly-L-lysine-coated coverslips (Sigma-Aldrich, UK) and incubated at 37˚C, 5% CO 2 for at least 1 hour. Patch pipettes of resistance 2-5 MV were pulled from thin-walled borosilicate glass tubing (Intracel, UK), fire-polished and coated with Sigmacote (Sigma-Aldrich, UK). An Axopatch 200B amplifier (Molecular Devices, CA, USA) was used for recordings which were filtered at 2 kHz and digitised at 5-10 kHz using a Digidata 1322A A/D converter (Molecular Devices, CA, USA). Cells were held at a potential V h , of 290 mV, with holding current less than 250 pA and series resistance less than 10 MV. Currents were recorded with cell capacitance compensated and leak currents subtracted using a P/4 subtraction protocol. Series resistance was compensated up to 80%. Standard current-voltage (I-V) protocols used 30-millisecond sweeps from 290 to +70 mV in 5 mV steps, from V h 290 mV. I-V curves were fitted with a Boltzmann function: I5G(V-V rev )/ 1+exp(-(V-V 50,act )/k), where V rev is reversal potential, V 50,act is voltage for half maximal current activation; G is conductance and k is slope factor. Conductancevoltage relationships were derived from individual I-V plots and fitted with a Boltzmann function: G/G max 51/1+exp((V 50,act -V)/k), where G max is maximal conductance. Peak conductance G, at each test potential was calculated from the corresponding peak current I, as follows: G5I/(V-V rev ). Steady-state inactivation protocols comprised a test pulse of +10 mV following 500-millisecond pre-pulse potentials between 2120 and +40 mV, from V h 290 mV. Peak current values were normalised to I max and curves fitted with the following Boltzmann function: I/I max 51/1+exp((V t -V 50,inact )/k), where I max is maximum peak current and V 50,inact , the midpoint of voltage-dependent inactivation. Resting membrane potential (E m ) was estimated by current clamp in physiological saline solutions (PSS), using highresistance patch pipettes (.10 MV). Extracellular PSS contained (in mM): NaCl 140, KCl 4, MgCl 2 1, CaCl 2 2, glucose 11, HEPES 10 adjusted to pH 7.4 with NaOH and 320 mOsm.litre 21 with sucrose. Internal PSS contained (in mM): Kglutamate 125, KCl 20, CaCl 2 0.37, MgCl 2 1, K-ATP 1, EGTA 1, HEPES 10 adjusted to pH 7.2 with KOH and 310 mOsm.litre 21 with sucrose. Data acquisition and analysis were performed using pCLAMP (v9.0, Molecular Devices) and Origin (v8.0, Microcal Software).

IonWorksH electrophysiology
High-throughput electrophysiology used the IonWorksH Quattro TM Automated Patch Clamp System (Molecular Devices, CA, USA). The extracellular solution was phosphate-buffered saline (pH 7.4). Internal solution contained (in mM): Kgluconate 100, KCl 40, MgCl 2 3.2, HEPES 10, EGTA 3, adjusted to pH 7.2 with KOH and 310 mOsm/litre with sucrose. Cells were resuspended in PBS at 1610 6 cells.ml 21 . Seal resistances ranged from a user-defined lower limit of 50 MV to ,250 MV. Data were rejected if the pre-compound amplitude of the Na + current was less than 50 pA or if the total resistance (parallel seal plus membrane resistance) decreased by more than 50% from pre-to post-compound measurement. Electrical access to the interior of cells was achieved using the pore-forming antibiotic amphotericin B (10 mg/100 ml internal solution). I Na sensitivity to TTX was determined by comparing peak I Na before and after a 2minute incubation with 1 mM TTX. All IonWorksH data are presented as maximum current amplitude (I max ). The IonWorksH platform does not provide cell capacitance measurements. Thus, current densities (pA.pF 21 ) cannot be calculated.

Quantitative PCR
Total RNA extraction was performed using the RNeasy kit (Qiagen, UK). RNA yield and purity were determined by spectrophotometry and only samples with an A 260 :A 280 ratio above 1.8 were used for experiments. Total RNA (1 mg) was reverse transcribed using the Precision TM Reverse Transcription Kit (PrimerDesign, UK) and qPCR performed using cDNA obtained from 25 ng of total RNA. qPCR was performed using an ABI 7500 Sequence Detection System with software version 1.2.3 (Applied-Biosystems, UK). All primers were specific for each gene of interest (Table 1). Amplification and detection were carried out in 96-well Optical Reaction Plates (Applied-Biosystems, UK) with SYBRH green Precision TM 26qPCR Mastermix (PrimerDesign, UK). All expression data were normalised to 18S rRNA expression, as determined by an initial reference gene screen using the geNorm TM Housekeeping Gene Selection Kit (PrimerDesign, UK). Primer-specificity was confirmed at the end of each qPCR run through the generation of single peaks in melt-curve analyses. Data analysis was performed using the 2 2DDCT method.

Western immunoblotting
Cells grown in 10-cm diameter Petri dishes were washed in PBS and lysed on ice in RIPA buffer with cOmplete Mini EDTA-free protease inhibitor cocktail (Roche, UK). Cell lysates were passed through a fine-gauge syringe needle 106to shear genomic DNA and centrifuged at 1000 g for 30 seconds. Protein samples were separated by 7-10% SDS-PAGE for 90 minutes at 160 V and transferred by electrophoresis (110 V for 90 minutes) onto a nitrocellulose membrane (Whatman, UK). Membranes were 'blocked' at room temperature (RT) for 1 hour in 5% (w/v) BSA in Tris-buffered saline (TBS) with 0.1% Tween-20 (TTBS), washed 36with TTBS and probed with primary antibody (1:1000-5000) in 5% BSA/TTBS at 4˚C overnight. Membranes were then re-washed with TTBS 36and incubated with appropriate HRP-conjugated secondary antibody (1:2000) in 5% BSA/TTBS at RT for 90 minutes. After further washing with TTBS, blots were treated with Western LightningH ECL reagent (PerkinElmer, UK) and immunoreactive proteins detected by exposure to film (GE Life Sciences, UK). In all cases, loading controls of bactin or total-protein (for phosphorylated proteins) were used in parallel.

Immunocytochemistry
Cells were seeded at 3610 4 onto sterile 0.01% poly-L-lysine-coated 22-mm square coverslips and incubated for 48 hours. After washing with PBS, cells were fixed with 4% (w/v) paraformaldehyde at RT for 20 minutes. Paraformaldehyde was quenched with 0.1 M glycine for 10 minutes; cells were permeabilised with 0.5% saponin (10 minutes; Sigma-Aldrich, UK) and blocked with 5% BSA/PBS for 1 hour. Cells were incubated at RT for 1 hour with primary antibody (1:200 in 5% BSA/PBS), washed 36 with PBS, incubated at RT for 1 hour with a FITCconjugated anti-rabbit secondary antibody (1:1000) in 5% BSA/PBS before final washing in PBS. Nuclei were stained with DAPI (100 ng.ml 21 ), prior to mounting coverslips onto glass slides with ProLongH Gold Antifade Reagent (Invitrogen, UK). Images were taken on a DeltaVision RT restoration microscope (Applied Precision, UK) using a 606oil immersion objective lens. Appropriate wavelength filters were used for FITC and DAPI fluorophores. Images were collected using a Coolsnap HQ camera (Photometrics, UK) with a Z optical spacing of 0.1 mm in 12 sections. Raw images were deconvolved using SoftWoRxH software and displayed as maximum projections using NIH Image J (W.S. Rasband, National Institutes of Health, USA). To confirm primary antibody specificity, wild-type and HEK-293 cells stably transfected to express Na v 1.7 were used as negative and positive controls, respectively. Non-specific binding of primary and secondary antibodies was assessed using the above protocol without either the primary antibody or secondary antibody incubation steps.

Immunohistochemistry
Cell lines (H460 and A549) and cancerous versus normal tissue (prostate and lung; Tristar Technologies LLC, MD, USA) were paraffin-embedded and cut into 4-mm sections. Sections were processed on a LabVision Autostainer 360 (Thermo Fisher Scientific, UK) using the ChemMate Envision TM + System-HRP (DAB+) kit (Dako, UK), according to manufacturer's instructions. Sections were incubated with anti-Na v 1.7 antibody (1:200) for 1 hour, followed by 30 minutes with HRPlabelled polymer-conjugated anti-rabbit IgG secondary antibody. Sections were counterstained with dilute Mayer's haematoxylin (Dako, UK). All processing was performed at RT. Images were taken with a light microscope and visual comparisons were made. Negative controls (without incubation of primary antibody) were performed for all experiments.

Proliferation
Cells were plated at 4000 cells per well into 96-well plates and cell numbers monitored in real time by in vitro micro-imaging using an IncuCyte TM incubator (Essen BioScience, UK), allowing for hourly monitoring of cell proliferation by determining cell confluence.

Migration
For scratch-wound assay, cells were seeded into 24-well culture plates at 2.5610 5 cells.ml 21 and then grown to 100% confluence. A linear wound was introduced in the centre of the cell monolayer. Culture medium supplemented with 1% FBS (to minimise cell proliferation) was then replaced and cell migration (wound closure) monitored using an IncuCyte TM incubator.

Invasion
Invasion was analysed in 24-well plates containing BD Matrigel TM Invasion Chambers (8-mm pore size polyethylene terephthalate membrane cell culture inserts covered with a film of Matrigel TM Basement Membrane Matrix; BD Biosciences, UK). The upper compartment was seeded with 4610 4 viable cells in basal culture medium. The lower compartment contained culture medium plus 20% FBS as a chemoattractant. After 48 hours at 37˚C, 5% CO 2 , inserts were washed in PBS for 2 minutes and cells fixed with ice-cold methanol for 10 minutes. Cells were then stained with haematoxylin (Sigma-Aldrich, UK) for 3 minutes and invaded cells (those attached to the lower side of the membrane) were counted in the whole insert using a light microscope at 2006magnification.
Adhesion 96-well plates were coated with 20 mg.ml 21 fibronectin at 37˚C for 1 hour, with some wells left uncoated as negative controls. After two washes with washing buffer (0.1% BSA in cell culture medium without serum), wells were 'blocked' with blocking buffer (0.5% BSA in cell culture medium without serum) at 37˚C, 5% CO 2 for 1 hour. After two washes, cells were plated at 20,000 cells per well in serum-free medium and incubated at 37˚C, 5% CO 2 for 30 minutes. Plates were left to shake at 2000 rpm for 30 seconds before being washed 36. Remaining cells were fixed with 4% paraformaldehyde at RT for 15 minutes, washed 36 and stained with Crystal Violet for 10 minutes. Wells were washed thoroughly with water and left to dry completely before addition of 2% SDS (30 minutes at RT). Absorbance of the resulting SDS/Crystal Violet solution was measured at 550 nm to quantify numbers of adherent cells.
For each functional assay, cells were seeded at the same densities for control (un-treated) and treated samples. Cell proliferation, migration and invasion data were then normalised as follows: cell proliferation to 0% confluence at 0 hours, migration to 0% wound confluence at 0 hours, and invasion to the control condition.

Generation of dilution clones
H460 cell dilution clones were generated using a serial dilution protocol provided by Corning, UK (http://www.level.com.tw/html/ezcatfiles/vipweb20/img/img/ 34963/3-2Single_cell_cloning_protocol.pdf). To increase the likelihood that cells of a particular colony originated from a single cell, selected cells were re-cloned for a second time. For all experiments, parental control H460 cells were also included to provide a direct comparison with the standard population of H460 cells. To confirm that dilution clones did not revert to their original state (maintenance of the Na v phenotype), I Na was routinely recorded by patch-clamp electrophysiology prior to any experiments.

Transient transfection
siRNA H460 cells were transfected with small interfering RNAs (siRNAs) directed against SCN9A (Na v 1.7; 59-TCGGATAGTGAATACAGCAAA-39), SCN1B (Na v b1; 59-CACATTGAGGTAGTGGACAAA-39) or control scrambled siRNAs (AllStars negative control siRNA; Qiagen, UK). Cells were seeded into 24-well plates (1610 5 ) in 0.5 ml basal culture medium. For each well, 37.5 ng of siRNA was diluted into 100 ml of medium without serum. HiPerFect transfection reagent (3 ml per well; Qiagen, UK) was added to diluted siRNA and mixed by brief vortexing. After 10-minute incubation at RT, transfection complexes were added drop-wise to cells giving a final siRNA concentration of 5 nM per well. Cells were incubated at 37˚C, 5% CO 2 for 24 hours before experiments were performed.
Fluorescence measurement of intracellular Na + Intracellular Na + concentration [Na + ] i , was measured using the ratiometric fluorescent dye SBFI as described by (Roger et al., 2007). Calibration of [Na + ] i was performed in each cell by comparing fluorescence in high (20 mM) and low (10 mM) Na + calibration solutions containing (in mM): NaCl 10 or 20, KCl 150, HEPES 5, EGTA 10, monensin 0.01, adjusted to pH 7.4 with NaOH and 320 mOsm/litre with sucrose.

Data analysis
All quantitative data are presented as the means 6 standard error of the mean (s.e.m.). Statistical analysis was carried out using Student's t-test or ANOVA (One-way with Student-Newman-Keuls post hoc correction), as appropriate (GraphPad Prism v5.0), with 95% confidence limits.