Voltage‐dependent activation of Rac1 by Nav1.5 channels promotes cell migration

Abstract Ion channels can regulate the plasma membrane potential (Vm) and cell migration as a result of altered ion flux. However, the mechanism by which Vm regulates motility remains unclear. Here, we show that the Nav1.5 sodium channel carries persistent inward Na+ current which depolarizes the resting Vm at the timescale of minutes. This Nav1.5‐dependent Vm depolarization increases Rac1 colocalization with phosphatidylserine, to which it is anchored at the leading edge of migrating cells, promoting Rac1 activation. A genetically encoded FRET biosensor of Rac1 activation shows that depolarization‐induced Rac1 activation results in acquisition of a motile phenotype. By identifying Nav1.5‐mediated Vm depolarization as a regulator of Rac1 activation, we link ionic and electrical signaling at the plasma membrane to small GTPase‐dependent cytoskeletal reorganization and cellular migration. We uncover a novel and unexpected mechanism for Rac1 activation, which fine tunes cell migration in response to ionic and/or electric field changes in the local microenvironment.

Na v 1.5 activity promotes motility and invasive behavior through several mechanisms, including MAPK activation (House et al., 2010(House et al., , 2015, upregulation of CD44 expression (Nelson, Yang, Millican-Slater et al., 2015), and potentiation of Na + /H + exchanger type 1 (NHE1) activity leading to cysteine cathepsin activation (Brisson et al., 2011;Brisson et al., 2013;Gillet et al., 2009). Na v 1.5 also promotes acquisition of an elongate mesenchymal-like phenotype, cortactin phosphorylation and cytoskeletal reorganization (Brisson et al., 2013;Nelson, Yang, Millican-Slater et al., 2015). Taken together, these data point to an important role for both Na v 1.5 and V m in regulating morphological changes and migration. An important question is, therefore, how migratory behavior is regulated by electrical changes and Na + flux at the plasma membrane.
Here, we investigated the role of Na v 1.5 in regulating V m in nonexcitable breast cancer cells. We report that Na v 1.5 promotes steady state V m depolarization, which in turn leads to localized activation of Rac1 at the plasma membrane, promoting morphological changes and migration. This study uncovers a novel and unexpected voltage-dependent mechanism for Rac1 activation, which would fine tune cell migration in response to ionic and/or electric field changes in the local microenvironment.

| Chemicals
Iberiotoxin, NS-1619, tetrodotoxin (TTX) and veratridine were purchased from Alomone Labs. Phenytoin was from Sigma. Ionomycin was from Cayman Chemical. EHT1864 was from Santa Cruz Biotechnology. Drugs were reconstituted as stock solutions according to manufacturer guidelines and diluted directly into culture medium and/or recording solutions at the indicated working concentrations.

| Whole-cell patch clamp recording
The whole-cell patch clamp technique was used to record membrane current and voltage from cells grown on glass coverslips for 48-72 hr.
Voltage-and current-clamp recordings were made using a Multiclamp 700B amplifier (Molecular Devices) at room temperature. Membrane currents and V m were digitized using a Digidata 1440A interface (Molecular Devices), low-pass filtered at 10 kHz and analyzed using pCLAMP 10.4 software (Molecular Devices). In voltage-clamp mode, signals were sampled at 50 kHz, and the series resistance was compensated by 40-60%. Linear components of leak were subtracted using a P/6 protocol (Armstrong & Bezanilla, 1977). The standard extracellular physiological saline solution (PSS) contained the following components (in mM): 144 NaCl, 5.4 KCl, 1 MgCl 2 , 2.5 CaCl 2 , 5 HEPES, and 5.6 glucose adjusted to pH 7.2 with NaOH. For the Na + -free PSS, NaCl was replaced with 144 mM N-methyl-D-glucamine (NMDG) or choline chloride (ChoCl) adjusted to pH 7.2 with HCl. The standard intracellular (pipette) solution contained (in mM) 5 NaCl, 145 KCl, 2 MgCl 2 , 1 CaCl 2 , 10 HEPES, and 11 EGTA adjusted to pH 7.4 with KOH. Free Ca 2+ concentration in the solution was calculated using MaxChelator software (Stanford University). Na + current was activated by depolarizing from −120 mV (250 ms) to voltages in the range −80 to +30 mV in 5 mV increments (50 ms).

| Membrane potential recording
The steady-state V m was recorded in I = 0 mode within 5 s of achieving the whole-cell configuration. To monitor the continuous V m in individual cells in response to pharmacological treatments, cells were held in the whole-cell configuration for 6 min, with 60 s in the where IC 50 is the concentration of drug at which half of its maximal effect occurs; and n is the slope giving the Hill coefficient. from −120 mV (250 ms) to voltages in the range −60 to +90 mV in 10 mV increments (300 ms). The outward current data were fitted to a single exponential decay (Sanguinetti & Jurkiewicz, 1990 where k is the rate constant.

| Intracellular Na + and Ca 2+ imaging
Measurement of [Na + ] i was performed as described in (Roger et al., 2007)

| Cell migration assay
Cell migration was measured using wound healing assays (Yang et al., 2012). Label-free ptychographic microscopy was used to monitor motility in real time (Marrison, Raty, Marriott, & O'Toole, 2013;Suman et al., 2016). Images were acquired over 16 hr at 9 min intervals using a Phasefocus VL-21 microscope with an 10× (0.25 NA) lens using 640 nm illumination and equipped with an environmental chamber maintaining the cells at 37°C in 5% CO 2 . The wound healing experiment was repeated three times on separate days. Image sequences of gap closure were processed using Phasefocus Cell Analysis Toolbox (CAT) software to segment and track individual cells at the leading edge and measure wound area. For each image sequence, the following parameters were automatically measured: change in normalized gap area over time; t 1/2 for gap closure (hr), determined by applying a linear fit to the normalized wound area reduction from t = 0 until the time at which the area reduces by 50%; collective migration (µm/hr), defined as: where slope is the rate of area of scratch closure and l is the length of the scratch; instantaneous velocity per cell (µm/s), considering segmented cells with track lengths of ≥ 5 frames; and directionality of leading-edge cells tracked for ≥ 5 frames, relative to the center of the scratch.

| Cell proliferation and invasion assays
Cell proliferation was measured using the thiazolyl blue tetrazolium bromide (MTT) assay, as described (Yang et al., 2012). Cell invasion was quantified using 24-well Corning BioCoat Matrigel Invasion Chambers according to the manufacturer's guidelines. Briefly, 2.5 × 10 4 MDA-MB-231 cells were seeded in the upper compartment supplemented with 1% FBS and appropriate treatments. The lower compartment contained 10% FBS and appropriate treatments. Cells were incubated at 37°C for 24 hr before the removal of non-invaded cells from the upper chamber using cotton buds. Cells that had invaded through the polyethylene terephthalate membrane were fixed using 4% paraformaldehyde for 10 min and were washed three times with phosphate buffered saline (PBS), 5 min/time, before staining with DAPI. The membrane was then mounted on a glass microscope slide. DAPI-positive cells on the membrane were viewed at 20× using a Nikon Eclipse TE200 epifluorescence microscope.
Experiments were performed on duplicate wells and repeated ≥ 3 times.

| Immunocytochemistry
Cells (1.6 × 10 5 ) were cultured on glass coverslips in 4-well plates for 48 hr to form confluent monolayers. Wounds were made on the coverslip using a P200 pipette tip and debris was washed off with 37°C DMEM. Appropriate treatments were subsequently applied, and cells were allowed to migrate into the wound for 3 hr. This incubation duration was used because it was long enough to show an effect of V m inhibition on migration. Cells were fixed in 4% paraformaldehyde for 10 min and immunocytochemistry was performed, as described (Nelson et al., 2014). The following primary antibodies were used: rabbit anti-Na v 1.5 (1:100; Alomone); rabbit anti-K Ca 1.1 (1:100; Neuromab); mouse anti-Rac1-GTP (1:500; New-East Biosciences); rabbit anti-total Rac1 (1:10; NewEast Biosciences); and mouse anti-CD44 (1:100; AbD Serotec). The secondary antibodies were Alexa 488 or 568-conjugated goat anti-mouse and Alexa 488 or 568-conjugated goat anti-rabbit (1:500; Life Technologies). In some experiments, cells were counter-stained with Alexa 633-conjugated phalloidin (1:50; Life Technologies). In experiments where permeabilized cells were labeled with Alexa 568-conjugated annexin V (1:20; Life Technologies), primary and secondary antibodies were incubated in Ca 2+ -containing annexin V binding buffer (Life Technologies) to preserve annexin V binding. Coverslips were then mounted on glass microscope slides using Prolong Gold with DAPI (Life Technologies). The slides were examined using a Zeiss LSM 710 confocal microscope at 40×. Confocal images were acquired sequentially for each channel. Images (512 × 512 or 1,024×1,024 pixels, dependent on experiment) were initially processed with the Zeiss Zen 2 software, and later exported into ImageJ for analysis.
Brightness/contrast was adjusted using the ImageJ "Auto" function.

| Morphology and lamellipodium scoring
Circularity was calculated on manually segmented cells using the free-hand tool in ImageJ (Schneider, Rasband, & Eliceiri, 2012): Values approaching 0 indicate a more elongated cell morphology.
Feret's diameter was measured as the maximum distance between any two points along the cell boundary. Three experimental repeats were performed, and 20-25 cells per condition were analyzed for every repeat. To score lamellipodia formation, cells were categorized into two groups: with and without lamellipodium. For each of the three experimental repeats, 40-50 cells per condition were analyzed blinded to treatment.

| Analysis of Rac1 localization in the lamellipodium
The signal densities of F-actin, Rac1-GTP and total Rac1 in the lamellipodium were analyzed in arcs at various distances from the leading edge as described (Dang et al., 2013). The region of interest in the lamellipodia was determined by the dense staining of F-actin.
To avoid bias in choosing ROIs, the analysis was carried out blinded to treatment. The ImageJ Radial Profile Extended plugin (Philippe Carl, CNRS, France) was used with starting radius = 0.43 µm, radius increment = 0.43 µm, ending radius = 5.6 µm, and total integrated angle = 90°, resulting in 13 arcs in total. Signal densities on each of the arcs were obtained and were normalized to those on the innermost arc. Typically, 20-30 cells at the edge of the wound were YANG ET AL.

| 3953
analyzed for each condition and the experiment was repeated three times.

| Colocalization analysis
Colocalization of annexin V and Rac1-GTP labeling within regions of interest drawn around the lamellipodium was evaluated using the Coloc2 plugin in ImageJ. The approximate point spread function on acquired images of 1,024 × 1,024 pixels was 2.71 pixels. For each region of interest, the thresholded Manders M1 and M2 coefficients for annexin V and Rac1-GTP were computed (Manders, Verbeek, & Aten, 1993), together with the Li's intensity correlation coefficient (Li et al., 2004). Measurements were taken from 30 cells per condition over three experimental repeats.

| In vitro Rac1 activity assay
Cells were cultured to 70% confluency in 6-well plates before addition of pharmacological treatment(s). After incubation for 24 hr, total cellular Rac1 activity was evaluated in cell lysates using a colorimetric ELISA-based small GTPase activation assay, according to the manufacturer's instructions (G-LISA; Cytoskeleton, Inc.; Antonov et al., 2012). Measurements were obtained from duplicate wells and the experiment was repeated three times. Forty-eight hours following transfection, appropriate treatments in phenol red-free DMEM were applied for 3 hr before imaging. FRET imaging of biosensor activation was performed at room temperature using a Nikon Eclipse TE200 epifluorescence microscope with a 40× (NA 0.60) objective. Acquisition from cells expressing low levels of the biosensor was performed at 8-bit depth using a Rolera-XR Fast 1394 CCD camera (QImaging) controlled by SimplePCI software (Hamamatsu). Acquisition time for donor and FRET channels was typically 200 ms. The donor and acceptor channels were acquired sequentially. The donor, monomeric teal fluorescent protein (mTFP), was excited at 436 nm and emission was collected at 480 nm. FRET emission was collected above 540 nm. The acceptor (Venus) was excited at 480 nm and emission collected above 540 nm. Visualization of cells transfected with a control plasmid expressing mTFP alone, created by site-directed mutagenesis (Nelson et al., 2014), revealed minimal bleed-through in the FRET and Venus channels.

| Live cell imaging of Rac1 biosensor activation
Images were taken from ≥ 49 cells across three experimental repeats.
Image processing was performed using ImageJ. Cells were manually segmented using the distribution of biosensor (donor mTFP signal).
Images were then background subtracted before calculation of the FRET/mTFP integrated density ratio for each cell.

| Statistics
Data are presented as mean ± standard error of mean (SEM). GraphPad Prism 7.0d was used to perform all statistical analyses. Polar histograms were plotted using OriginLab Origin 2017. For normally distributed data, paired or unpaired Student's two-tailed t test was used to compare two groups, as appropriate. Multiple comparisons were made using one-way analysis of variance (ANOVA) or repeated measures ANOVA followed by Tukey post hoc tests, as appropriate. For repeated measures data that were not normally distributed, a Friedman test with Dunn's post hoc tests was used. Fisher's exact test was used to examine the distribution of samples within a population. Contingency table data were analyzed using χ 2 test. Colocalization data were analyzed by two-way ANOVA. Results were considered significant at p < .05.

| RESULTS
3.1 | Na v 1.5 channels depolarize the membrane potential of breast carcinoma cells The V m of motile cancer cells is generally more depolarized than in terminally-differentiated noncancer cells (Yang & Brackenbury, 2013). In addition, both depolarized V m and Na v 1.5 channel activity correlate with metastatic potential (Fraser et al., 2005). However, it is not known whether the relationship between Na v 1.5, V m and cell behavior is causal.
Here, we first investigated whether Na + conductance through plasma membrane Na v 1.5 channels contributes to the depolarized V m that has been reported previously in MDA-MB-231 cells (Fraser et al., 2005).
Na v 1.5 carries a persistent inward Na + current, small in amplitude compared with the transient Na + current, which plays an important role in shaping the action potential firing pattern, especially in the subthreshold voltage range (George, 2005). In nonexcitable cells, this would allow significant accumulation of Na + over an extended period, potentially permitting V m depolarization at steady state. Na v 1.5 is robustly expressed in MDA-MB-231 cells, both in intracellular compartments and at the plasma membrane ( Figure 1a). This agrees with previous reports of VGSC distribution in nonexcitable cells (Brisson et al., 2011(Brisson et al., , 2013Carrithers et al., 2007Carrithers et al., , 2009Fraser et al., 2005;Nelson, Yang, Millican-Slater et al., 2015;Persson et al., 2014;Yang et al., 2012). We measured Na + currents carried by Na v 1.5 using whole-cell voltage-clamp recording. Depolarization from −120 mV to −10 mV elicited both transient and persistent inward Na + currents (Figure 1b,c). Analysis of the voltage-dependent activation and steady-state inactivation of the Na + current revealed a window current between −50 mV and −10 mV ( Figure   1d,e). We next investigated whether Na v 1.5 could regulate the V m . Stable suppression of Na v 1.5 expression using lentiviral shRNA (Nelson, Yang, Millican-Slater et al., 2015) significantly hyperpolarized the V m compared with control cells, measured using whole-cell current-clamp recording, from −12.6 ± 0.9 mV to −16.0 ± 0.8 mV (p < .01; n = 16; t test; Figure 1f).
Thus, Na v 1.5 is required to depolarize the steady-state V m .
We next evaluated the effect of pharmacologically perturbing Na v 1.5 activity. The specific VGSC blocker TTX (30 µM) reversibly inhibited the transient and persistent Na + current (p < .001 and p < .05, respectively; F I G U R E 1 Na v 1.5 is endogenously expressed in breast carcinoma cells and regulates the membrane potential. (a) MDA-MB-231 cells labeled with Na v 1.5 antibody (green), phalloidin to label the actin cytoskeleton (red), and DAPI to label the nucleus (blue). Insets show cell peripheries at higher magnification. (b) Typical whole-cell recording showing large transient and small persistent Na + current. The cell was depolarized to −10 mV for 50 ms following a 250 ms pre-pulse at −120 mV. (c) Expanded view of persistent Na + current 40-45 ms following onset of depolarization. (d) Activation and steady-state inactivation of Na + current. Normalized conductance (G/G Max ) was calculated from the current data and plotted as a function of voltage (n = 14). Normalized current (I/I Max ) was plotted as a function of the pre-pulse voltage (n = 9). Data are fitted with Boltzmann functions. (e) Expanded view of shaded area under activation and inactivation traces highlighting the window current. (f) Effect of Na v 1.5 shRNA knock-down on the V m compared to nontargeting shRNA control (n = 16). Data are mean and SEM. *p < .05; **p < .01; Student's t test. DAPI, 4',6-diamidino-2-phenylindole; shRNA, short hairpin RNA n ≥ 5; repeated measures ANOVA with Tukey test; Figure 2a- consistent with previous reports (Brackenbury et al., 2007;Fraser et al., 2005;Roger et al., 2003). TTX also significantly hyperpolarized the V m from −13.2 ± 1.3 mV to −17.5 ± 1.6 mV (p < .01; n = 17; repeated measures ANOVA with Tukey test; Figure 2e,f). On the other hand, the carrier for TTX (148 µM sodium citrate, pH = 4.8) had no effect on the V m ( Figure S1a,b). In agreement with the TTX data, the VGSC-inhibiting antiepileptic drug, phenytoin (100 µM) also significantly reduced the transient and persistent Na + current and hyperpolarized the V m ( Figure   S1C-K). In contrast, veratridine (100 µM), which increases channel open probability by inhibiting inactivation (Ulbricht, 1998), increased the persistent Na + current (Figure 2g,h; Figure S2a,b) and depolarized the V m from −16.37 ± 1.4 mV to −12.6 ± 1.3 mV (p < .05; n = 13; t test; Figure 2h).
In summary, these data show that pharmacological or genetic inhibition of Na v 1.5 hyperpolarizes the steady-state V m , whereas increasing the persistent Na + current depolarizes the V m . Thus, persistent current carried by Na v 1.5 contributes to V m depolarization.

| Membrane potential depolarization is dependent on extracellular Na +
To evaluate the sufficiency of extracellular Na + to determine the V m , we next removed Na + in the extracellular solution during recording.
Using NMDG to replace extracellular Na + had a similar hyperpolarizing effect ( Figure S2c,d). Thus, extracellular Na + is required for steady-state V m depolarization.
Inward Na + current through Na v 1.5 channels would be expected to elevate the intracellular Na + level ([Na + ] i ]) at steady state. Indeed, VGSCs have previously been shown to increase [Na + ] i (Campbell, Main, & Fitzgerald, 2013;Roger et al., 2007). We therefore next investigated the involvement of extracellular Na + in setting [Na + ] i using the ratiometric Na + indicator SBFI-AM (Figure 3c,d). Replacement of extracellular Na + with NMDG significantly reduced the steady-state [Na + ] i from 15.6 ± 1.7 mM to 7.6 ± 2.9 mM (p < .001; n = 3; ANOVA with Tukey test; Figure 3e). Treatment with TTX also significantly reduced the steady-state [Na + ] i , suggesting that Na v 1.5 activity contributes to this elevation of [Na + ] i (p < .05; n ≥ 10; ANOVA with Tukey test; Figure 3e). In agreement with this, the [Na + ] i of Na v 1.5-shRNA cells was significantly lower than for control cells (9.4 ± 1.2 mM vs.13.5 ± 0.6 mM; p < .01; n = 3; t test; Figure 3f). On the other hand, Na v 1.5 inhibition with TTX had no effect on [Ca 2+ ] i ( Figure S3a-d). Together, these data suggest that Na + influx through Na v 1.5 channels increases [Na + ] i , but not [Ca 2+ ] i .
3.3 | K Ca 1.1 regulates the membrane potential but not the Na + level Na v 1.5 channels enhance cell migration, invasion, tumor growth, and metastasis (Fraser et al., 2005;Roger et al., 2003).
Given that Na v 1.5 depolarizes the steady-state V m , and that V m depolarization is functionally instructive in regulating cellular behavior (Yang & Brackenbury, 2013), we reasoned that Na v 1.5 may regulate cell migration via setting the V m . Thus, we sought a means by which to manipulate the V m independent of Na v 1.5 to separate the effects of V m and [Na + ] i on downstream signaling. To do this, we took advantage of the ubiquitously expressed large conductance Ca 2+ -activated K + channel, K Ca 1.1 (Contreras et al., 2013). We verified that K Ca 1.1 was robustly expressed in MDA-MB-231 cells (Khaitan et al., 2009;Ma et al., 2012;Roger, Potier, Vandier, Le Guennec, & Besson, 2004), and its expression was unaffected by Na v 1.5 downregulation (Figure 4a,b). Next, using the perforated patch clamp mode to maintain endogenous [Ca 2+ ] i , we found that the K Ca 1.1 opener NS-1619 (1 µM; Macmillan, Sheridan, Chilvers, & Patmore, 1995) increased outward K + current, confirming K Ca 1.1 activity 3.4 | Na v 1.5-dependent membrane potential depolarization regulates cell migration and morphology Using NS-1619 and TTX as tools to modify the V m independent of, or as a result of Na v 1.5 activity, respectively, we next investigated the effect of V m depolarization on cell migration in wound healing assays using labelfree ptychographic imaging (Figure 5a). Both TTX and NS-1619 slowed the rate of wound closure (Figure 5b). The t 1/2 of wound closure increased from 5.7 ± 1.1 hr to 9.8 ± 0.7 hr in the presence of TTX and 9.9 ± 0.1 hr in the presence of NS-1619 (p < .05; n ≥ 5; ANOVA with Tukey test; Figure 5c). Similarly, the collective migration of cells moving into the wound was reduced from 7.8 ± 0.7 µm/hr to 5.5 ± 0.4 µm/hr in the presence of TTX and 4.9 ± 0.5 µm/hr in the presence of NS-1619 (p < .05; n ≥ 5; ANOVA with Tukey test; Figure 5d). The instantaneous velocity of individual cells at the leading edge was also reduced from 6.2 × 10 −3 ± 5.6 × 10 −5 µm/s to 4.7 × 10 −3 ± 5.0 × 10 −5 µm/s and 4.4 × 10 −3 ± 4.5 × 10 −5 µm/s, for TTX-and NS-1619-treated cells, respectively (p < .001; n ≥ 2,662; Kruskal-Wallis with Dunn's test; Figure 5e). In addition, both TTX and NS-1619 caused a small, but statistically F I G U R E 2 Tetrodotoxin hyperpolarizes, and veratridine depolarizes, the membrane potential. (a) Representative trace showing the inhibitory effect of tetrodotoxin (TTX; 30 µM) on Na + current, and recovery after washout. The cell was held at −120 mV for 250 ms before depolarizing to −10 mV for 50 ms.  Figure 5f). Hyperpolarizing the V m with NS-1619 did not significantly alter cell proliferation measured in an MTT assay ( Figure S4a). Given that inhibiting Na v 1.5 with TTX has no effect on the proliferation of MDA-MB-231 cells (Fraser et al., 2005;Gillet et al., 2009;Roger et al., 2003), together, these data suggest that both Na v 1.5 activity and V m depolarization promote cellular migration, without affecting proliferation. On the other hand, whilst TTX significantly reduced invasion into Matrigel, consistent with previous reports (Brackenbury et al., 2007;Driffort et al., 2014;Fraser et al., 2005;Roger et al., 2003;Yang et al., 2012), NS-1619 had no effect ( Figure S4b,c). Thus, Na v 1.5 promotes cellular migration via V m depolarization, whereas Na v 1.5-dependent promotion of invasion occurs via a V m -independent mechanism.
Na v 1.5 promotes an elongate, mesenchymal-like motile morphology (Brisson et al., 2013;Driffort et al., 2014;. This morphological modulation has been shown to occur through two F I G U R E 3 Extracellular Na + sets the membrane potential and intracellular Na + level. (a) V m in control physiological saline solution, after extracellular Na + replacement with choline chloride and following washout. Solid line, mean; gray shading, SEM (n = 10). (b) Quantification of V m over the last 5 s in control, choline chloride, and washout (n = 10). (c) Representative SBFI fluorescence intensity (ratio of emission at 340 nm/380 nm) when cells were perfused with standard physiological saline (center panel), solution containing 10 mM Na + and 20 μM gramicidin (left panel) and solution containing 20 mM Na + and 20 μM gramicidin (right panel). Ratio images are color-coded so that warm and cold colors represent high and low [Na + ], respectively. Scale bar = 10 μm.  (Nelson et al., 2014, and (b) via increased Src kinase activity, phosphorylation of the actin-nucleation-promoting factor cortactin, and F-actin polymerization (Brisson et al., 2013). In addition, the V m has been shown to promote reorganization of the actin filament network and cytoskeleton in other cell types (Chifflet et al., 2003(Chifflet et al., , 2004(Chifflet et al., , 2005Nin et al., 2009;Szaszi et al., 2005).
However, the potential involvement of V m in Na v 1.5-mediated signaling is not known. We thus investigated the participation of Na v 1.5-mediated V m depolarization in regulating cell morphology.
Given that Na v 1.5-mediated V m depolarization promotes both morphological changes and cellular migration, we postulated that it might regulate formation of lamellipodia in migrating cells. We therefore scored the number of migrating cells with visible lamellipodia in a wound healing assay following treatment with TTX and NS-1619. Both treatments significantly reduced the proportion of cells with a lamellipodium in the measured population (p < .001; n ≥ 139 cells per condition; χ 2 test; Figure 6d). Together, these findings suggest that Na v 1.5-mediated V m depolarization promotes a change in cellular morphology towards an elongate phenotype, and increases lamellipodia formation, thus potentiating cellular migration.
3.5 | Na v 1.5-dependent membrane potential depolarization regulates Rac1 activation The small GTPase Rac1, together with Rho and cdc42, plays a critical role in regulating cytoskeletal organization and cell motility (Burridge & Wennerberg, 2004). In addition, Rac1 also regulates the formation of lamellipodia, cell motility and the directionality of cell movement (Wu et al., 2009). We therefore postulated that Na v 1.5 and V m depolarization may regulate the level of active (GTP-bound) Rac1 in the lamellipodia of migrating cells. We treated migrating cells with NS-1619 and TTX for 3 hr and then evaluated distribution of Rac1-GTP by quantifying immunocytochemical signal density across concentric arcs in the lamellipodium of individual cells (Figure   7a,b). As reported previously, a large quantity of Rac1 is present intracellularly in the perinuclear region, although there is also an enrichment of Rac1-GTP in the lamellipodium, consistent with its critical role in this region (Garcia-Mata et al., 2011;Wu et al., 2009).
Both TTX and NS-1619 significantly reduced the peak Rac1-GTP signal at the leading edge of migrating cells (p < .05; n ≥ 66; one-way ANOVA with Tukey test; Figure 7c). However, the level of total cellular Rac1-GTP, determined in whole-cell lysates, was unaffected by TTX treatment (p = .80; n = 6; t test; Figure 7d). Given that the majority of Rac1 is present intracellularly (Figure 7a; Das et al., 2015;Moissoglu, Slepchenko, Meller, Horwitz, & Schwartz, 2006), these data suggest that Na v 1.5 may regulate Rac1 locally at the plasma membrane and/or within lamellipodia. We next evaluated the effect of TTX and NS-1619 on distribution of Rac1-GTP versus total Rac1 using an antibody that does not distinguish between GDP-and GTPbound forms of Rac1 (Figure 7e). The peak total Rac1 signal at the leading edge of migrating cells was unaffected by both TTX and NS-1619 (p = .69; n ≥ 60; one-way ANOVA; Figure 7f,g). Thus, the ratio of peak Rac1-GTP to total Rac1 at the leading edge was significantly reduced by both TTX and NS-1619 (p < .001; n = 3; one-way ANOVA; Figure 7h). These data suggest that V m depolarization caused by steady state Na v 1.5 activity increases Rac1 activation at the leading edge of migrating cells.
V m depolarization promotes clustering and activation of the small GTPase K-Ras as a result of voltage-dependent redistribution of charged phospholipids including phosphatidylserine (Zhou et al., 2015). Given that Rac1 is also anchored to the inner leaflet of the plasma membrane via interaction with phosphatidylserine (Finkielstein et al., 2006;Magalhaes & Glogauer, 2010), we reasoned that a similar voltage-dependent mechanism may be responsible for Rac1 activation in the lamellipodia. We therefore next evaluated the effect of V m depolarization on colocalization of Rac1-GTP with phosphatidylserine, using annexin V as a marker (Figure 8a). Treatment of migrating cells with TTX and NS-1619 reduced colocalization of Rac1-GTP with phosphatidylserine in the lamellipodia (Figure 8b-d).
Similarly, the TTX and NS-1619 treatments both reduced the Li's F I G U R E 4 The large conductance Ca 2+ -activated K + channel K Ca 1.1 regulates the membrane potential but not intracellular Na + . (a) MDA-MB-231 cells labeled with K Ca 1.1 antibody (green), phalloidin to label the actin cytoskeleton (red), and DAPI to label the nucleus (blue). (b) Western blot of K Ca 1.1 in control MDA-MB-231 cells and cells in which Na v 1.5 has been knocked down with shRNA. Positive control = rat brain lysate. Loading control = α-tubulin. (c) Representative perforated patch clamp recording showing activation of outward current using the K Ca 1.1 activator (NS-1619; 1 µM) and inhibition with iberiotoxin (100 nM). The cell was held at −120 mV for 250 ms before depolarization to + 60 mV for 300 ms. (d) Current-voltage relationship of the K Ca 1.1 current. Cells were held at −120 mV for 250 ms before depolarization to voltages ranging from −60 to +90 mV in 10 mV steps for 300 ms (n = 5). Data are fitted with single exponential functions. To evaluate whether Na v 1.5-mediated V m depolarization promotes Rac1 activity in live cells, and to independently confirm the immunocytochemistry data, we next employed a genetically encoded Rac1 FRET biosensor to monitor Rac1 activation. In agreement with a previous report (Fritz et al., 2015), the biosensor showed a general gradient of Rac1 activation increasing from the cell interior towards the periphery, consistent with its critical role in actin remodeling at the edge of migrating cells (Figure 9a; Wu et al., 2009). Treatment with TTX did not affect biosensor distribution (Figure 9a), but suppressed activation at the periphery (Figure 9b,c). Quantification of FRET across the whole cell revealed that TTX significantly reduced Rac1 activation to 48.6 ± 5.2% of control (p < .001; n ≥ 49; t test; Figure 9d). Together, these results indicate that blockade of Na v 1.5 channels suppresses Rac1 activation at the cell periphery.

| 3963
EHT1864 increased circularity and reduced Feret's diameter in a dosedependent manner (Figure S5a,b). Furthermore, EHT1864 increased circularity to a similar extent to TTX (p < .001; n ≥ 308; one-way ANOVA with Tukey test; Figure 10b,c). Importantly, co-application of EHT1864 with TTX had no additive effect on circularity or Feret's diameter (Figure 10b,c). These results show that Rac1 activity is required to transduce the Na v 1.5-dependent V m depolarizing signal to promote acquisition of an elongate, motile cell phenotype (Figure 10d).

| DISCUSSION
Here we identify Na v 1.5-mediated V m depolarization as a regulator of Rac1 activation. Thus, we link ionic and electrical signaling at the plasma membrane to small GTPase-dependent cytoskeletal reorganization and cellular migration. We show that depletion of extracellular Na + or blockage of Na v 1.5 channels reversibly hyperpolarizes the V m at the timescale of minutes. We further show that Na v 1.5-dependent V m depolarization increases Rac1 colocalization with phosphatidylserine at the leading edge of migrating cells, promoting Rac1 activation, and resulting in acquisition of a motile, mesenchymal-like cellular phenotype. We therefore propose that Na v 1.5 may serve as a sensor for local changes in the ionic microenvironment, thus permitting voltage-dependent activation of Rac1 to fine tune cell migration.

| Ion conductance, membrane potential, and migration
We show that Na v 1.5 carries a persistent inward Na + current in MDA-MB-231 cells. In agreement with the work of others, we report incomplete inactivation of Na v 1.5 resulting in a small voltage window encompassing the V m , thus supporting this persistent Na + current (Brisson et al., 2011;Djamgoz & Onkal, 2013;Gillet et al., 2009;Roger et al., 2003;Yang et al., 2012). The persistent inward Na + current carried by Na v 1.5 contributes to steady-state V m depolarization. Our findings agree with another study in H460 non-small cell lung cancer cells, in which Na v 1.7 channels were shown to depolarize the V m by~10 mV (Campbell et al., 2013). Thus, the mechanism identified here is likely applicable to other nonexcitable cell types where VGSCs are expressed (Chen et al., 2019;Grimes et al., 1995;House et al., 2010;Persson et al., 2014;Zhou et al., 2015). In addition, it has previously been shown that VGSCs contribute to steady-state V m depolarization and viability in rat optic nerve axons and spinal cord astrocytes over a time course of hours (Sontheimer, Fernandez-Marques, Ullrich, Pappas, & Waxman, 1994;Stys, Sontheimer, Ransom, & Waxman, 1993), suggesting that this mechanism persists over an extended period. There may be other Na + -permeable pathway(s) also contributing to V m regulation in nonexcitable cells, for example, via epithelial Na + channels (Amara, Ivy, Myles, & Tiriveedhi, 2016), Na + -K + ATPase (Winnicka et al., 2008), and NHE1 (Brisson et al., 2011). The V m may also be dependent on the activity of transporters and channels regulating the movement of other ions, including Cl − and K + (Yang & Brackenbury, 2013). The depolarized resting V m reported here may therefore be attributed not only to the high permeability to Na + but also to the low expression/activity of hyperpolarizing K + channels.
Relatively small alterations in V m can be functionally significant. For example, during synaptic transfer from photoreceptors to ganglion cells in turtle retina, V m depolarizations as small as 5 mV are sufficient to activate paired ganglion cells (Baylor & Fettiplace, 1977). The V m can function as an instructive signal to regulate cell cycle progression in proliferating cells (Cervera et al., 2016;Cone, 1971;Cone & Cone, 1976;Zhou et al., 2015). V m depolarization can regulate other cellular behaviors, including differentiation (Sundelacruz et al., 2008) and cytoskeletal reorganization (Chifflet et al., 2003(Chifflet et al., , 2004(Chifflet et al., , 2005Nin et al., 2009;Szaszi et al., 2005). At the tissue level, changes in V m can also regulate morphogenesis, regeneration and tumorigenesis (Beane et al., 2011(Beane et al., , 2013Cervera et al., 2016;Chernet & Levin, 2014;Chernet et al., 2016;Lobikin et al., 2012Lobikin et al., , 2015. Several studies have shown that epithelial cells undergo V m depolarization when migrating into wounds (Chifflet et al., 2004(Chifflet et al., , 2005. However, an interesting and novel finding of our study here is that V m depolarization is not simply a consequence of motile behavior (Schwab et al., 2012), but is itself a master regulator of morphological changes and migration. Here, we found that hyperpolarizing the V m by~5 mV reduced migration by~30%. The fact that this level of V m hyperpolarization did not completely inhibit migration raises the possibility that additional hyperpolarization may have a further inhibitory effect. It is also likely that other ion conductance routes contributing to, and/or dependent on the V m , may be involved in migration regulation. For F I G U R E 7 Na v 1.5 and V m regulate Rac1 activation/distribution. (a) Images of representative cells after treatment with TTX (30 µM) and NS-1619 (1 µM) for 3 hr. Cells were labeled with Rac1-GTP antibody (green), phalloidin (red), and DAPI (blue). Arrows in the Rac1-GTP panels highlight the distribution or lack of expression at the leading edge. (b) Rac1-GTP signal density, measured across 20 arcs, in 0.43 µm radius increments, within a quadrant mask region of interest at the leading edge, normalized to the first arc (n ≥ 66). (c) Peak Rac1-GTP signal density per cell from (b), normalized to the first arc (n ≥ 66). (d) Total Rac1-GTP quantified in whole-cell lysates using colorimetric small GTPase activation assay (n = 6). (e) Images of representative cells after treatment with TTX (30 µM) and NS-1619 (1 µM) for 3 hr. Cells were labeled with Rac1-GTP antibody (green), total Rac1 antibody (red), and DAPI (blue). Arrows in the Rac1-GTP panels highlight the distribution or lack of expression at the leading edge. (f) Total Rac1 signal density, measured across 20 arcs, in 0.43 µm radius increments, within a quadrant mask region of interest at the leading edge, normalized to the first arc (n ≥ 59). (g) Peak Rac1 signal density per cell from (f), normalized to the first arc (n ≥ 59). (h) Ratio of Peak Rac1-GTP/Peak total Rac1 for each experimental repeat, normalized to control (n = 3). Data are mean and SEM. *p < .05; **p < .01; ***p < .001; ANOVA with Tukey test. ANOVA, analysis of variance; DAPI, 4',6-diamidino-2-phenylindole; SEM, standard error of mean; TTX, tetrodotoxin example, hyperpolarization would increase the driving force for Ca 2+ entry. Nonetheless, our data underscore that persistent Na + influx through Na v 1.5 channels leads to V m depolarization, which in turn promotes cell migration.
Given the general trend towards depolarized V m both in cancer cells (Cervera et al., 2016;Fraser et al., 2005;Yang & Brackenbury, 2013), and in epithelial cells undergoing migration (Chifflet et al., 2004(Chifflet et al., , 2005, we propose that voltage-dependent migratory behavior may be a general cellular phenomenon. For example, although involvement of V m was not directly investigated, activation of the V mhyperpolarizing channel K Ca 1.1 with NS-1619 reduces migration of glioma cells (Kraft et al., 2003). Furthermore, VGSCs have been detected in cells from a broad range of tumor types, where they potentiate migration and invasion (Besson et al., 2015; F I G U R E 9 Na v 1.5 regulates Rac1 activation in live cells detected using a genetically encoded Rac1 FRET biosensor. (a) Images of representative cells after treatment ± TTX (30 µM) for 3 hr. Rac1 activation biosensor distribution is shown in the donor (mTFP) channel. Images are color-coded so that warm and cold colors represent high and low values, respectively, for sensor distribution and activation (FRET). (b) Fluorescence intensity profile along line drawn across control cell in (a). (c) Fluorescence intensity profile along line drawn across TTX-treated cell in (a). (d) Emission ratio (FRET), for cells measured after treatment ± TTX (30 µM) for 3 hr, normalized to control (n ≥ 33). Data are mean and SEM. **p < .01; Student's t test. TTX, tetrodotoxin (1 µM) for 3 hr (n = 30). (f) Li's intensity correlation quotient for Rac1-GTP and annexin V colocalization (n = 30). Data are mean and SEM. **p < .01; ***p < .001; ANOVA with Tukey test. ANOVA, analysis of variance; DAPI, 4',6-diamidino-2-phenylindole; SEM, standard error of mean; TTX, tetrodotoxin F I G U R E 1 0 Na v 1.5-mediated morphological changes and migration are dependent on Rac1 activation. (a) Images of representative cells after treatment with TTX (30 µM) ± Rac1 inhibitor (EHT1864; 0.5 µM) for 3 hr. Cells are labeled with CD44 antibody (red). (b) Circularity of cells after treatment with TTX (30 µM) ± EHT1864 (0.5 µM) for 3 hr (n ≥ 308). (c) Feret's diameter (µm) of cells after treatment with TTX (30 µM) ± EHT1864 (0.5 µM) for 3 hr (n ≥ 308). Data are mean and SEM. ***p < .001; ANOVA with Tukey test. (d) Proposed mechanism underlying Na v 1.5-mediated V m -dependent morphological changes and migration. Na v 1.5 channels carry Na + influx, which depolarizes the V m , causing redistribution of charged phosphatidylserine in the inner leaflet of the phospholipid bilayer, promoting Rac1 redistribution and activation. Rac1 regulates cytoskeletal modification via the Arp2/3 complex and increasing phosphorylation of cortactin and cofilin, promoting acquisition of a promigratory phenotype (Pollard, 2007;Stock & Schwab, 2015). Na + influx through Na v 1.5 channels may also impact on migration and invasion through other mechanism(s), including via β1 subunit-dependent adhesion (Brisson et al., 2013;House et al., 2010House et al., , 2015Nelson et al., 2014). ANOVA, analysis of variance; TTX, tetrodotoxin 2012). We therefore argue that the results presented here have broad applicability to other cell types.

| Voltage-dependent regulation of Rac1 activation
V m depolarization promotes K-Ras clustering and subsequent activation (Zhou et al., 2015). Various small GTPases, including K-Ras and the Rho GTPases, are anchored to the inner leaflet of the plasma membrane through interaction with the anionic phospholipids PIP2, PIP3, and phosphatidylserine (Finkielstein et al., 2006;Heo et al., 2006;Magalhaes & Glogauer, 2010;Remorino et al., 2017;Zhou et al., 2015). In the case of K-Ras, clustering and activation has been shown using electron microscopy to arise as a result of depolarization-induced nanoscale redistribution of phosphatidylserine and PIP2 in the inner leaflet of the phospholipid bilayer, to which the K-Ras is anchored (Zhou et al., 2015).
Our major discovery here is that Rac1 activation is also voltagedependent. Our data show that Na v 1.5-mediated V m depolarization causes Rac1 activation, likely as a result of voltage-dependent redistribution of phosphatidylserine ( Figure 10d). Thus, V m -dependent signaling is not exclusively limited to mitogenic cascades but is also a key regulator of morphological changes and migration.
A growing body of evidence implicates V m depolarization in regulation of small GTPase activity. For example, V m has been shown to regulate the activity of GEF-H1, which, in turn, regulates the activity of the Rho/Rho kinase pathway (Waheed et al., 2010). V m depolarization-dependent activation of Rho and subsequent cytoskeletal organization has been shown to be Ca 2+ -independent (Szaszi et al., 2005). In agreement with this, our data also imply that Ca 2+ is not involved as a signaling intermediary in V m depolarization-induced Rac1-mediated promotion of migration. On the other hand, in ATPstimulated microglia, VGSC activity increases [Ca 2+ ] i , and activates ERK and Rac1, promoting migration (Persson et al., 2014). V m depolarization-induced activation of Ras and Rap1 has also been reported in mouse cortical neurons (Baldassa, Zippel, & Sturani, 2003).
The mechanism uncovered here may feed into, and interact with, additional signaling pathways which have been shown to be dependent on Na v 1.5 activity. For example, Na v 1.5 has been shown to promote Src kinase activity and phosphorylation of cortactin and cofilin (Brisson et al., 2013). Thus, Na v 1.5-induced V m depolarization may activate Rac1, increasing cortactin phosphorylation and therefore enhancing cofilin activity and actin filament polymerization. Src has also been shown to regulate Rac1 activity, suggesting the potential for feedback regulation (Servitja, Marinissen, Sodhi, Bustelo, & Gutkind, 2003). Na v 1.5 positively regulates the expression of the metastasispromoting protein CD44, which may activate Src and therefore contribute to this process (Bourguignon, Zhu, Shao, & Chen, 2001;House et al., 2010House et al., , 2015Lee, Wang, Sudhir, & Chen, 2008;Nelson, Yang, Millican-Slater et al., 2015). In addition, Na v 1.5 is expressed on astrocytes where it regulates migration by promoting reverse Na + / Ca 2+ exchange (Black, Dib-Hajj, Cohen, Hinson, & Waxman, 1998;Pappalardo, Samad, Black, & Waxman, 2014). Furthermore, Na v 1.5 is expressed on the endosome of macrophages, where it regulates endosomal acidification (Carrithers et al., 2007). On the other hand, Na v 1.6 is expressed within macrophage podosomes, where it promotes invasion via intracellular Na + release (Carrithers et al., 2009). Na v 1.5 and Na v 1.6 are also expressed on microglia, where the latter has been shown to also regulate migration and cytokine release (Black, Liu, & Waxman, 2009). It remains unknown whether Na v 1.5 opening changes during migration and what the signals might be that would regulate such alterations; however, it has been shown that Na v 1.5 can be regulated by growth factors and hormone signaling, and is itself mechanosensitive (Beyder et al., 2010;Fraser et al., 2014). Together, these studies suggest that VGSCs may further regulate the behavior of nonexcitable cells via pathways in addition to the one identified here.

| Implications for our understanding of metastasis
We provide evidence that Na v 1.5 contributes to steady-state V m depolarization, which in turn promotes V m -dependent Rac1 activation, lamellipodial protrusion formation and cellular migration. These data are in agreement with the previously reported role for VGSCs in regulating morphology and migration of tumor cells (Brackenbury & Djamgoz, 2006;Brackenbury et al., 2007;Brisson et al., 2013;Fraser, Ding, Liu, Foster, & Djamgoz, 1999;Fraser et al., 2005;. It is therefore conceivable that depolarization-dependent Rac1 activation may contribute to metastatic dissemination in response to local ionic changes in the tumor microenvironment. Thus, pharmacological targeting of VGSCs, which inhibits metastasis in preclinical models (Driffort et al., 2014;Yildirim, Altun, Gumushan, Patel, & Djamgoz, 2012), may provide therapeutic benefit via V m hyperpolarization and downregulation of V m -dependent small GTPase activation. In conclusion, our results reveal a new role for Na v 1.5 channels as voltage-dependent activators of Rac1 signaling to promote cellular migration.

ACKNOWLEDGMENTS
This study was supported by the Medical Research Council (G1000508), Cancer Research UK (A25922), and a Radhika V Sreedhar Scholarship.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

CONFLICT OF INTERESTS
The authors declare that there are no conflict of interests.

AUTHOR CONTRIBUTIONS
M. Y. and W. B. contributed to the conception and design of the work.
O., T., and W. B. contributed to drafting the work and revising it critically for important intellectual content. All authors approved the final version of the manuscript.