Electrophoresis of cellular membrane components creates the directional cue guiding keratocyte galvanotaxis

Summary

Background

Motile cells exposed to an external direct current electric field will reorient and migrate along the direction of the electric potential in a process known as galvanotaxis. The underlying physical mechanism that allows a cell to sense an electric field is unknown, although several plausible hypotheses have been proposed. In this work we evaluate the validity of each of these mechanisms.

Results

We find that the directional motile response of fish epidermal cells to the cathode in an electric field does not require extracellular sodium or potassium, is insensitive to membrane potential, and also insensitive to perturbation of calcium, sodium, hydrogen, or chloride ion transport across the plasma membrane. Cells migrate in the direction of applied forces from laminar fluid flow, but reversal of electroosmotic flow did not affect the galvanotactic response. Galvanotaxis fails when extracellular pH is below 6, which suggests that the effective charge of membrane components may be a crucial factor. Slowing the migration of membrane components with an increase in aqueous viscosity slows the kinetics of the galvanotactic response. In addition inhibition of PI3K reverses the cell’s response to the anode, suggesting the existence of multiple signaling pathways downstream of the galvanotactic signal.

Conclusions

Our results are most consistent with the hypothesis that electrophoretic redistribution of membrane components of the motile cell is the primary physical mechanism for motile cells to sense an electric field. This chemical polarization of the cellular membrane is then transduced by intracellular signaling pathways canonical to chemotaxis to dictate the cell’s direction of travel.

Introduction

For over a century it has been known that motile cells exposed to external information from an applied direct current electric field will migrate along the orientation of the electrical potential (galvanotax/electrotax) [1]. Cells respond to currents that are similar in magnitude to those that exist under normal physiological conditions, including during the development of embryos of some animals [2] and wound formation [3] due to a short-circuit of the trans-epithelial potential [4]. In addition, exogenous electric fields applied in-vivo are sufficient to disrupt development [5] or produce directed migration [6]. At this time the mechanisms that cells use to sense an external electrical field, transduce this signal to the cell migration apparatus, and then appropriately change the direction of migration remain controversial.

Galvanotactic behavior has been demonstrated thus far in over thirty metazoan-derived cell types including neurons [7], lung cancer cells [8], and leukocytes [9] as well as in crawling single celled organisms including Dictyostelium discoideum [10] and many swimming (ciliated) protozoa [11]. It is far less common to see reports of animal cells that fail to galvanotax and this usually correlates with poorly motile behavior [6]. Electric fields that produce galvanotaxis are typically in the range of 0.1 to 10 V/cm [3]. It has been established that galvanotaxis operates independently of sensing an external chemical gradient [12], therefore we can limit our discussion of a cellular sensor of an external electric field to the electrical dimensions of the cell.

These electrical properties of the cell are primarily dictated by the cell’s plasma membrane. External to the plasma membrane, the cell adheres to a charged substrate and is bathed by a conductive ionic media. Due to the high resistance of the cellular plasma membrane compared to the external media as well as the small size of the cell, most (≳ 99.999%) of the current flow created by an external electric field will pass around the cell and will therefore have limited effect on intracellular components [13]. The shielding effect of the plasma membrane is bridged primarily by a set of membrane channels with selective permeability to ions. In addition, the plasma membrane itself is embedded with a large set of charged macromolecules and lipids, which will be directly acted on by an external electric field through Coulombic interactions. These extracellular charged components and the charged substrate will also induce electro-osmotic flow in the presence of an external electric field.

Given these physical constraints we can limit our exploration of the galvanotactic sensing mechanism to the following set of four plausible physical hypotheses (Figure 1). (A) Cells will be asymmetrically excited due to hyperpolarization of the anodal side and depolarization of the cathodal side of the cell, changing the opening probability of voltage gated ion channels as well as creating an asymmetric electro-motive force for ionic flow once ion channels are open [10]. (B) Electro-osmotic flow created at the substrate will re-orient cells through hydrodynamic shear as is seen with laminar fluid flow [14]. (C) Electrostatic and electro-osmotic forces at the plasma membrane will apply mechanical force on the cell or on tension sensitive cell surface components. (D) These same electrostatic and electro-osmotic forces at the plasma membrane will also redistribute the charged components of the membrane establishing a cathodal/anodal axis of polarity [15]. These non-exclusive mechanisms are summarized in Figure 1.

Figure 1. Models for directional sensing of a keratocyte in an electric field.

Visual description of the possible models for a galvanotactic response of a motile cell. (A) An electric field will polarize the cell changing electro-motive forces and opening/closing voltage gated ion channels. (B) Electro-osmotic flow, ν_eo, at the charged migration surface will apply external force on the cell, which could for instance displace adhesions laterally. (C) Electro-osmotic forces created by the relatively immobile charged ions in the cell membrane attracting a mobile double layer at the cell surface will combine with electro-static forces on charged macromolecules and membrane components to produce mechanical work. As depicted, this could asymmetrically activate a force sensor creating a local signal that could be used to define the front and the back of the cell. (D) Local electro-osmotic and electrostatic forces at the cell membrane will also electrophorese membrane components. Negatively charged components will move to the anode, positively charged components will migrate to the cathode. Electro-osmotic forces at the membrane will also act to push proteins to one side of the cell or the other depending on the net surface charge of the cell.

Each of these putative sensors of an external electric field would require signal transduction pathways to relay the directional information to the cytoskeletal players that produce cell migration. Most cell types respond to an electric field by migrating towards the cathode, although some (often similar) cell types respond by migrating to the anode [16, 17]. The mechanism underlying these anti-parallel responses is unclear. Separate reports on the same cell type (human polymorphonuclear cells) have found opposing anodal versus cathodal galvanotactic responses [6, 18], which has been attributed to differences in extracellular calcium [19]. In addition, a mutant strain of Dictyostelium has been identified with a reversed (anodal) electrotactic response. This mutant phenotype could be replicated by inhibiting both cGMP and PI3K signaling activity [20], supporting the hypothesis that there is a separation between the physical mechanism of sensing an electric field and the eventual directional response.

Downstream of the unknown sensing mechanism, the current literature supports a hypothesis where intracellular signaling pathways canonical to chemotaxis are used to transduce the galvanotactic signal. It is commonly noted that inhibition of PI3K disrupts the galvanotactic response of cells [21]. Galvanotaxis can also be blocked by inhibition of alternative signaling pathways such as VEGF, ERK and Rho/ROCK [16, 22]. In addition cells in electric fields have asymmetric distributions of common polarity factors including phosphatidylinositol (3,4,5)-trisphosphate (PIP3), PTEN, and growth factor receptors [9, 22, 23]. However, the signal transduction pathways of chemotaxis and galvanotaxis do not completely overlap, as unlike chemotaxis PTEN inhibition improves the strength of a cell’s galvanotactic response [9, 24], and in general the signaling pathways of galvanotaxis remain poorly understood.

The final step in the directional response to an electric field is the actual change in organization of the cytoskeleton of the motile cell to produce a change in direction. Little is known about the mechanical requirements for this process other than a described independence from the microtubule system and a general requirement for actin polymerization [25].

In this work, we seek to identify the cellular sensor of an external electric field by investigating the validity of each of the hypothetical physical mechanisms that could produce a galvanotactic response using the motile fish epithelial keratocyte model system. Keratocytes move at high speeds, with a simple shape, and, unlike cultures of mammalian cells, are robust to extreme physical perturbations, making them useful for understanding mechanical effectors of motility. In addition, keratocytes operate largely without requirements for external stimuli and are not known to be chemotactic. We find that the most likely sensing mechanism for galvanotaxis occurs due to electrophoretic redistribution of membrane components to the anode of the cell defining the rear. This polarity of membrane components is transduced by canonical intracellular signalling pathways that then dictate the cell’s directionality.

Results

Keratocytes migrate to the cathode in an electric field

To examine the motion of spontaneously motile keratocytes, cells were imaged in sets of ~600 μm wide fields of view every minute for one hour. Cells under control conditions were equally likely to move in all directions (Figure 2.a). Application of a 10 V/cm direct current electric field biased the motion of motile cells towards the cathode (Figure 2.b). Populations of cells exhibited a dose response in directional bias to applied potential, with a statistically significant response at 0.25 V/cm and a fully saturated response at 3 V/cm (Figure 2.c).

Figure 2. Keratocytes migrate to the cathode in an electric field.

(a) Rose plots of the distribution of angles traveled in populations of 137 control cells (p=0.56) and (b) 110 cells in an electric field of 10 V/cm (p = 5.9 × 10⁻⁹), with the cathode oriented toward the right. The p value is calculated from a Kolmogorov-Smirnov test for a uniform distribution of angles traveled. (c) The strength of the directional response is calculated for populations of cells as the mean ± standard error of the cos(θ), where θ represents the direction that a cell travels relative to the electric field lines, as depicted graphically. A cos(θ) of 1 indicates a complete directional response towards the cathode (purple line), a cos(θ) of 0 indicates no response (cyan line), and a cos(θ) of −1 indicates a reversed response. Green points and fit line represent the dose response to the applied potential of cells in normal media. For a given applied potential, there was a decreased strength of response when media conductivity was decreased by mixing L-15 media 1:4 (red) or 1:10 (blue) with water. (d) Re-plotting the same data as in panel c in terms of current flow shows that for all salt concentrations the directional response is proportional to the ionic current. Given the constant flow cell geometry used in these analysis, current density, J, will depend on the measured current, I, and cross-sectional area of the flow cell, A, where $J=\frac{I}{A}=\frac{I}{2\times {10}^{-7}{m}^2}$.

For a given potential drop, decreasing the density of ions flowing over the cells by decreasing the salt concentration of the media decreased the effectiveness of the galvanotactic response despite a constant applied field strength (Figure 2.c). We found that the strength of the galvanotactic response of cells in dilute and normal media collapsed to a single dose-response curve based on the dose of the ionic current density that flows over the cells (Figure 2.d).

Electric fields have previously been claimed to provide a kinetic cue as well as a directional cue, with speed increasing with increasing applied voltage [23, 26]. Similarly, under our standard conditions for inducing galvanotaxis within a small-volume chamber, we found that keratocyte cell speeds increased with applied potential. However, a significant increase in temperature due to resistive heating was measured, which will independently increase cell speed [27] (Figure S1). Thus in the robustly spontaneously motile keratocytes there is no evidence of an electro-kinetic effect, and as far as we can measure, we find that the electric field acts only to re-orient existing cell motility machinery.

Disruption of ionic flux across the cell membrane does not disrupt galvanotaxis

Galvanotaxing cells in an external electric field will experience an estimated ~1 to 10% asymmetry in membrane voltage polarization, (Supplementary Materials I), which in turn could produce asymmetries in ion flux through the plasma membrane and possibly provide a directional cue, (Figure 1.a). We examined this initial hypothesis by observing keratocyte migration in salt solutions without Na⁺ (replaced with K⁺ or Cs⁺) or without K⁺. Keratocyte migration remained intact under both salt conditions and cells continued to migrate towards the cathode (Figure 3). Similarly, removal of extracellular calcium has been reported to not modify galvanotaxis in keratocytes and fibroblasts [14, 28].

Figure 3. Ionic flux does not drive galvanotaxis.

Direction of travel of cells as quantified by the cos(θ) (blue) and mean cell speed (red) for populations of cells under specified perturbations, with n indicating number of cells analyzed and error bars indicating the standard error of the mean. All perturbations were performed under an electric field of 5 V/cm (1.5 mA) except for the no electric-field control cells. Perturbations include Na⁺ free salt solution (Na⁺ ions replaced with K⁺ ions) [similar results were seen with Cs⁺ supplementation], K⁺ free media (Na⁺ supplemented), Ca⁺⁺ ionophore A-21387, intracellular Ca⁺⁺ chelator BAPTA-AM, L-type voltage gated Ca⁺⁺ channel blocker verapamil, epithelial sodium channel (ENaC) and Na⁺/H⁺ exchanger (NHE-1) inhibitor amiloride, volume regulated anion channel (VRAC) inhibitors DCPIB and quinine, ATP and ADP (which can act as chloride channel inhibitors), vacuolar-type H⁺-ATPase inhibitor bafilomycin, and proton ionophore dinitrophenol.

Moreover, we found that the galvanotactic response of cells was insensitive to a Ca²⁺ ionophore as well as an intracellular calcium chelator and buffering agent (10 μM A-23187 and 10 μM BAPTA-AM). Inhibitors for L-type Ca²⁺ channels (50 μM verapamil), the Na⁺/H⁺ exchanger NHE1 (20 μM amiloride), or the volume regulated anion channel (50 μM DCPIB or 10 mM ATP or 10 mM ADP) again had no effect on the strength of cell’s galvanotactic response (Figure 3).

These perturbations will not only disrupt typical chemical gradients that exist across the cell membrane, but will also modify the membrane potential of the whole cell. The opening probability of a typical voltage gated ion channel depends on membrane potential over a range of ~20 to 60 mV. Therefore, replacement of Na⁺ ions in the media with K⁺ ions (expected depolarization, ϕ _M ≳ 0mV) or addition of calcium ionophore (expected hyperpolarization, ϕ _M ≳ −100mV , [29]) should completely abrogate the ability of the channel to respond to a 5 mV change in potential from an external electric field. Nevertheless, these drastic perturbations had no effect on galvanotaxis.

In addition to asymmetric flux of electrolytes, an intracellular pH (pH_i) gradient could be produced by an external electric field [30] and guide the direction of migration [31]. However, using the membrane permeable pH sensitive dye BCECF we found no detectable gradient of pH_i inside of either spontaneously motile cells or cells undergoing galvanotaxis (data not shown), and neither the H⁺-ATPase inhibitor bafilomycin nor the H⁺-ionophore dinitrophenol inhibited galvanotaxis (Figure 3).

Thus, after removal of any of the three most prevalent cations, repeated mechanistically distinct disruptions of the membrane potential that provides the electromotive driving force across the membrane, disruption of several sets of ion channels, and direct measurement of pH_i, we must conclude that there is no evidence to support the first hypothesis that asymmetries in ionic current through the plasma membrane drives the directional sensing of an electric field for a galvanotactic response.

Laminar fluid flow but not bulk electro-osmotic fluid flow can direct cell motility

A second hypothetical cellular sensor of an electric field that has been suggested previously [14] would be a mechanical cellular response to the electro-osmotic fluid flow created at the charged surface that the cell migrates on, (Figure 1.b, Supplementary Material II). In support of this hypothesis, we found that cells exposed to fluid forces from laminar flow reorient and migrate in the direction of fluid flow at shear stresses of around 2.5 Pa and above (Figure S2, Movie S1,S2).

We then directly tested the role of electro-osmotic flow in re-orienting cells in an electric field by reversing the direction of the electro-osmotic fluid flow. Under control conditions using negatively charged glass or tissue culture plastic as a substrate, electro-osmotic fluid flow at the surface was oriented toward the cathode with a magnitude of ~5 μm/s, (Figure 4.a). Following coating of the substrate with positively charge poly-L-lysine the direction of flow was reversed toward the anode (Figure 4.a). However, cells exposed to anodal electro-osmotic fluid flow did not modify their cathodal direction of motility or the sensitivity of the response (Figure 4.b).

Figure 4. Keratocytes migrate in the direction of applied force, but bulk electro-osmotic flow does not drive galvanotaxis.

(c) Measured flow profile seen when a 5 V/cm electric field exists across the flow cell as measured by fluorescent tracer particles. Negligible particle motion is noted when the electric field is off. Re-circulation flow in the center of the flow cell was noted due to the static head of pressure at the ends of the flow cell. Control conditions (left, red) produced flow toward the cathode. Coating surfaces with 20 μg/mL poly-l-lysine (right, blue) reversed the direction of electro-osmotic flow to the anode. Cartoon keratocyte is not drawn to scale. (d) Rose plot of directions of travel of keratocytes under an electric field of 1 V/cm (0.33 mA) with the cathode to the right under control conditions (left) and with the substrate patterned with 20 μg/mL poly-l-lysine (right). Cells exhibited a robust galvanotactic response to the cathode under both conditions.

Since the force created by electro-osmotic flow on the cell is far smaller than that created by laminar shear stress (Supplementary Material II) and reversal of the direction of flow did not reverse the direction of galvanotaxis we can conclude that our second hypothetical electric field sensor (electro-osmotic fluid flow arising from the substrate) is not the driving physical mechanism behind galvanotaxis.

Galvanotaxis is sensitive to changes in extracellular pH

As the plasma membrane electrically shields the interior of the cell, and ionic flow through ion channels does not produce the galvanotactic response, the ionic current flowing over cells must control cell direction by affecting parts of the cell that extend outside of the plasma membrane. An electric field will have two primary effects on these exterior cellular components. All charged components will experience an electro-static force, while both charged and uncharged components will experience drag force from electro-osmotic fluid flow along the cell surface [28, 32], (Supplementary Material III).

To assess the importance of these forces to sensing an external electric field, we modified the strength and possibly direction of the applied forces by changing the charge of the extracellular membrane components by modifying extracellular pH. We found that keratocytes could survive and retain motility over a surprisingly broad pH range of 5.3 to 9.5 (Movie S3). However, the ability of cells to respond directionally to an electric field, showed a dramatic dependence on pH (Figure 5.a). From a complete galvanotactic response at a pH of 6.2 (cos(θ) = 0.82) there was no measured response of cells at a pH of 5.8 (cos(θ) = −0.12). We can infer a direct effect of pH on the galvanotactic sensor as cell migration is otherwise stable (with stable cell speed), inhibition of galvanotaxis by an acidic pH was a reversible phenotype, and an acidic pH did not eliminate the ability of cells to respond directionally to shear stress (Figure S2). Of note, a pH of 6.0 has been previously identified in granulocytes as an isoelectric point in the switch between anodal migration in basic pH and cathodal migration in acidic pH [33].

Figure 5. Keratocyte galvanotaxis is sensitive to protonation.

(a) Measured cos(θ) (green) and cell speeds (magenta) ± standard error for cells in media of variable pH when exposed to an electric field of 5 V/cm. Cells retained robust motility across this pH range. All measurements were done without the presence of serum, except pH of 5.8, 6.2 and 7.4. The presence of serum did not effect directionality at an acidic pH. (b) The mean ± standard error of the electrophoretic mobility of red blood cells and keratocytes in suspension and in an electric field of 1.5 V/cm. Red blood cells phoresed towards the anode as would be expected from a negatively charged cell, keratocytes phoresed towards the cathode as would be expected from a positively charged cell. Note the electrophoretic mobility speed is an order of magnitude larger than typical speeds of cell migration.

Given that galvanotaxis fails critically as pH transitions from 6.2 to 5.8, the failure is most likely due to protonation. To distinguish the relative importance of electro-osmotic and electro-phoretic forces, the net charge of the cell at physiological pH of 7.4 was determined by measuring the electrophoretic mobility of cells in suspension [34]. Unlike red blood cells, keratocytes were found to have a net positive surface charge at a pH of 7.4 (Figure 5.b). Thus we can rule out a significant effect of electro-osmotic flow, as the surface charge of the cell is positive with electro-osmotic flow oriented toward the anode, and further protonation would only increase the zeta potential at the membrane (ζ_m), increasing the strength of electro-osmotic flow to the anode. Instead, this data supports a hypothesis where an electric field applies force to a net negatively charged membrane component toward the anode, defining the rear of the cell.

The kinetics of galvanotaxis are dependent on aqueous viscosity

The electrophoretic force applied to a net negatively charged membrane component in an external electric field could act as a galvanotactic sensor either through direct mechanotransduction (Figure 1.c) or by re-distributing components along the plasma membrane (Figure 1.d). The force applied per molecule in an electric field of 1 V/cm would be quite small, requiring 62 elemental charges to generate 1 fN of force. There are few reports in the literature of stretch activated channels or mechanical transduction pathways responding to forces in this femtonewton range, with typical stimuli being 1 to 10 pN or higher [35]. We also found that inhibition of stretch activated calcium channels with 100 μM gadolinium had no impact on cellular galvanotaxis (cos(θ) = 0.91 at 5 V/cm). However, it would only take 0.4 fN of force or 25 elemental charges to induce a significantly skewed protein distribution, plausible for highly glycosylated membrane proteins such as syndecans as well as stable oligomers (Supplementary Materials IV).

To distinguish between these two hypotheses, we examined the kinetics of individual cell’s directional response to an electric field, which for redistribution of membrane components will depend on the electrophoretic mobility. Theoretically we would predict that time required for the electrophoretic redistribution of charged membrane components and consequently the time for a cell to reorient in an electric field (time to respond) would both depend on the applied potential and the aqueous viscosity, while the time for a cell to lose directional orientation after an electric field is turned off (time to forget) would depend on the aqueous viscosity and the degree of previous polarization (Supplementary Materials IV). We found that the time time to respond was dependent on the strength of the applied potential (Figure 6). In addition we found that increasing the aqueous viscosity from 1 to 50 cP using methylcellulose increased both the time to respond and the time to forget. Increasing aqueous viscosity had no effect on cell speed (95% of control) and no obvious visual perturbation of cell migration.

Figure 6. Time to Respond and Time to Forget are dependent on aqueous viscosity.

The calculated mean(cos(θ)) is shown at every time point for all observed cells. Dashed black line represents control cells at steady state in the electric field, dashed blue line represents control cells at steady state without an electric field, red line indicates cells in 1 cP media and green line represents cells in 50 cP media + methyl-cellulose. (a) Graphical depiction of the time of cells to respond to a 1 V/cm (0.3 mA) electric field, where the electric field is turned on at the 15^th minute for cells in 1 and 50 cP media. We see cells in the lower viscosity media reach steady state behavior (red circle) faster than cells in higher viscosity media (green circle). (b) Graphical depiction of the time of cells to forget a 5 V/cm (1.5 mA) electric field, where the electric field is turned off at the 30^th minute for cells in 1 and 50 cP. Again we see cells in the lower viscosity media reach the new steady state behavior (red circle) faster than cells in higher viscosity media (green circle). (c) The mean ± standard error of the time to respond and time to forget were quantified from the time it takes each cell to reach steady state in minutes. The p-value of unpaired Student’s two-sample t-tests between normal and elevated viscosity are marked in red for each.

Confirming this result, we found that the time for a cell to switch directions in an electric field of 1 V/cm that was reversed had a weak but statistically significant positive correlation to cell size (Figure S3). These results are consistent with the hypothesis that electrophoretic motion of a membrane component to the anode/rear side of the cell dictates cell directionality. We directly visualized the redistribution of charged membrane components to the anode in an external electric field, with the fluorescently labeled lectin Concavalin A (ConA) (Figure S4). Re-distribution to the anode in an electric field of 10 V/cm was observed over a time scale of 1 to 3 minutes, comparable to the time required for cells to begin to alter their direction.

Starting from our original four hypothetical mechanisms for a cell to sense an external electric field (asymmetric ionic polarization, shear stress from bulk electro-osmotic flow, activation of a mechanical force sensor, and electrophoretic membrane component redistribution), our data directly supports only the final mechanism. This indicates that a cell senses an electric field by responding to the electrophoretic polarization of the components of its plasma membrane.

Downstream signaling pathways are required for transduction of the galvanotactic signal

A model of electrophoretically induced polarization of membrane components holds a great deal of similarity to models of chemotaxis. Where chemotaxis involves a non-uniform chemical environment that is interpreted through a uniform set of membrane receptors [36], galvanotaxis appears to involve a uniform chemical environment that is interpreted through a non-uniform set of membrane components. The net effect is to create an internal chemical “compass” in a motile cell, which is typically although not exclusively, represented as an asymmetric accumulation of PIP3 at the leading edge [37]. However, in the spontaneously motile and non-chemotactic keratocyte, we found no requirement for extracellular serum factors (Figure S5), suggesting that the electrophoresed component is constitutively active. Therefore to assess the identity of signal transduction pathways that define the internal compass, we investigated pharmacological inhibitors of known chemotaxis signaling pathways.

For keratocytes, we found that disruption of PIP3 production with the addition of PI3K inhibitor LY294002, greatly decreased the number of cells that were spontaneously motile, from 36% to 16% . Cells that retained motility moved at speeds that were ~75% of controls. Inhibition of PI3K activity did eliminate the directional response of keratocytes at a field strength of 1 V/cm and produced a strikingly reversed response, toward the anode, in a field strength of 5 V/cm (Figure 7). At both field strengths it does appear that the majority of cells are still oriented along the electric field lines; however, there is a dramatic increase in the number of cells with a reversed response, suggesting the existence of two competing signal transduction pathways downstream of the electrophoretic sensing mechanism.

Figure 7. Keratocyte galvanotaxis is sensitive to PI3K activity.

Rose plots of the distribution of angles traveled in populations of cells exposed to electric field of 1 and 5 V/cm (cathode oriented to the right) with and without the presence of PI3K inhibitor LY-294002. The fraction of cells traveling to the either the quartile of angles representing the cathode or anode are represented in red. Inhibition of PI3K causes some cells at 1 V/cm and a majority of cells at 5 V/cm to to reverse direction to the anode.

We used further pharmacological perturbations to gain a glimpse into possible roles of other signaling pathways. We found that disruption of PKC signaling with 50 μM clomiphene or 10 μM tamoxifen inhibited the directional response in cells to an electric field (cos(θ) = −0.3 and 0.59 respectively); however inhibition of the Rho-associated serine/threonine kinase with 25 μM Y-27632 had no effect on galvanotaxis (Figure S5).

Cytoskeletal reorganization produced by galvanotaxis resembles pathways used in spontaneous migration

The final step in the directional response of a motile cell to an external electric field is the reorganization of the cytoskeleton to change the direction of travel. We therefore wished to determine whether galvanotaxis altered or merely reoriented the autonomous motility machinery.

At steady state, we found that cells imaged in an electric field did not have a dramatically different appearance than cells imaged under control conditions (Figure S6, Movie S5). When the direction of the field was alternated, by reversing the polarity of the two electrodes, cells would switch directions to migrate toward the new cathode over a 5 to 10 minute time interval. Cells could change direction by either reversing their polarity (42% of events) or by smoothly turning in space (58% of events) (Figure S6). Cells that underwent a smooth turn in space developed asymmetries in shape that mirror those seen in keratocytes undergoing spontaneous turns in the absence of an electric field (Allen et al. in prep.). In addition, similar to granulocytes [18], keratocytes were often noted to periodically overshoot a straight path towards the cathode, producing path oscillations (Movie S6).

We have found that inhibition of non-muscle myosin II using the small molecule inhibitor blebbistatin [38] did not affect the directional response of keratocytes (cos(θ) = 0.98 at 5 V/cm). In addition, inhibition of adhesion maturation and signaling using the small molecule inhibitors of focal adhesion kinase FAK-14 [39] and PF-228 [40] similarly did not alter the ability of keratocytes to respond to an electric field (cos(θ) = 0.93 at 5 V/cm and cos(θ) = 0.87 at 1 V/cm respectively). Thus the electric field appears to establish an internal compass that can use endogenous mechanical pathways to cause cell turning, but does not have an absolute requirement for myosin contractility or for adhesion signalling and maturation.

Discussion

This work suggests that epithelial keratocytes re-orient in an electric field due to re-distribution of negatively charged membrane components to the anode, which is interpreted by intracellular signaling pathways commonly identified to play a role in chemotaxis to activate the same machinery that the cell uses for spontaneous turning. Re-distribution of components of the cell’s plasma membrane by an electric field has been experimentally demonstrated previously for ConA, low density lipoprotein receptor, epithelial growth factor receptors, fibronectin receptor, and acetylcholine receptor [28, 32, 41, 42], lending plausbility to this proposed mechanism.

The membrane is composed of a complex set of charged proteins, lipids and carbohydrates. For each charged membrane component exposed to an external electric field, there will be varying degrees of re-distribution dependent on the relative charge of the macromolecule and the cell membrane. The identity of a single critical macromolecule for sensing an electric field is unknown. However, we can put likely constraints on the identity of this hypothetical sensor as a mobile complex with a large net negative charge (>~ 25 e⁻), a critical change in net charge around a pH of 6.0 and a role in determining the orientation of migration. One hypothetical sensor would be the transmembrane heparan sulphate proteoglycan, Syndecan-4, that is thought to interact with the cytoskeleton to promote migration [43], including through PKC dependent pathways [44], and is found to be highly expressed in motile zebrafish keratocytes (S. Lou, unpublished data).

Downstream of this polarity in membrane components there is a clear role for intracellular signaling pathways, particularly in establishing cathodal versus anodal migration. In the work of Sun et al. (2012) [45] in keratocytes, the signal transduction pathway from the galvanotactic sensor to the machinery of cell motility is modeled as a strong PI3K dependent pathway that defines the front at the cathode and a weak ROCK dependent pathway that defines the back at the cathode. This is consistent with our data, where without PI3K activity at sufficiently high field strengths the secondary ROCK dependent pathway is capable of defining the cell rear at the cathode and reorienting cells to the anode. We additionally identified that PKC is critical to both pathways, consistent with an established, though not definitive, pro-migratory role [46].

Consequent to activation of these intracellular signaling pathways, we found that cells can use the same mechanical mechanisms for changing the direction of migration as cells migrating outside of an electric field. This is the first evidence that downstream signaling pathways, canonical to chemotaxis, not only exist in the spontaneously motile and non-chemotactic keratocyte model system but also are able to act as an internal compass when provided with an external directional cue from an electric field.

Conclusions

The sensing mechanism for the galvanotaxis of motile keratocytes most consistent with our data is the global electrophoretic redistribution of one or several membrane components carrying a sufficiently large net charge to overcome thermal noise. Specifically, it appears that a negatively charged membrane component is electrophoresed to the cell rear to initiate cell reorientation. This physical separation of membrane components is then transduced by at least two competing intracellular signaling pathways, including one dependent on PI3K. These signaling pathways then influence the otherwise autonomously acting machinery of cell motility to change the direction of cell migration. Our data rules out directional sensing from electro-osmotic fluid flow and argues against asymmetric transmembrane potential acting as the galvanotaxis sensor.

Experimental Procedures

In brief, experiments were performed with cultures of Central American cichlid, Hyposophys nicaraguensis, keratocytes exposed to DC electric fields in thin flow cells where single cell paths were measured by time-lapse phasecontrast microscopy. Fluid flow was measured by velocity of tracer particles and cell electrophoretic mobility was measured by velocity of tracer cells in suspension. Temperature was measured using rhodamine intensity, pHi by ratiometric dye BCECF, and the ConA distribution by Texas-red conjugated ConA. Full experimental details are provided in the Supplemental Experimental Procedures.

Highlights.

• We find no evidence of asymmetric ion flow into the cell as the galvanotactic sensor

• Cells move in the direction of fluid flow, but this is not required in galvanotaxis

• The charge and mobility of membrane components are critical to galvanotaxis

• A PI3K dependent pathway exists that transduces this polarization to the cytoskeleton