Effects of Direct Current Electric Fields on Cell Migration and Actin Filament Distribution in Bovine Vascular Endothelial CellsLi X. · Kolega J.
Department of Pathology and Anatomical Sciences, University of Buffalo School of Medicine and Biomedical Sciences, Buffalo, N.Y., USA Corresponding Author
Electric fields exceeding 1 V/cm occur during wound healing, morphogenesis, and tumor growth, and such fields have been shown to induce directional migration of a variety of different cells. However, the mechanism by which electric fields direct cell movement is not yet understood, and the effects on vascular endothelial cells are entirely unknown. We demonstrate that cultured bovine aortic endothelial cells migrate toward the cathode of an applied electric field. Time-lapse microscopic imaging shows that the field suppresses protrusive activity from anode-facing surfaces of the cells while stimulating protrusions from surfaces that face the cathode. The threshold for this response is 1–2 V/cm, similar to field strengths measured in vivo. In addition, fluorescence microscopy shows that lamellipodia projecting toward the cathode are rich in actin filaments. Using quantitative image analysis, we show that the electric field induces a transient 80% increase in the amount of filamentous actin in the cell. Comparison of the distribution of F-actin with total protein distribution indicates that F-actin is asymmetrically distributed in the cytoplasm, being selectively enriched toward the cathode. We propose that physiological electric fields direct cell migration by eliciting an intracellular signal that creates new sites for actin assembly in the cathodal cytoplasm.
Copyright © 2002 S. Karger AG, Basel
Because endothelial migration determines the rate and pattern of new vessel outgrowth, its control is important wherever one wishes to stimulate or inhibit angiogenesis. Examples include not only tissue repair, but also diabetic blindness, solid tumor growth and metastasis, and coronary heart disease . Most attention has been focused on the role of diffusible chemical signals such as vascular endothelial growth factor and other angiogenic materials. However, a potential alternative control is through the ability of electric fields to orient cell migration.
Electric fields of 1–2 V/cm have been shown to orient the movements of a wide variety of cells in vitro [2, 3, 4]. Because similar electrical potentials occur across wounds in mammalian skin  and cornea , and even larger electric fields (up to 5 V/cm) have been measured in developing limb buds , it has been proposed that endogenous, trans-tissue electric fields play a significant role in directing cell movements during vertebrate development and wound repair [reviewed in ref. 2]. Application of an external voltage to counter these endogenous fields can inhibit normal cell migration in amphibian limb buds , while supplementing the endogenous field accelerates it . In mammalian tissues, electrical stimulation has been shown to promote re-epithelization of epidermal wounds  and re-growth of nerve axons across experimentally severed spinal cords . Clinically, electric fields have been used for many years to induce closure of chronic skin ulcers  and to stimulate healing of bone fractures  even though the precise mechanism for this effect is still not understood. If endothelial cells respond to physiological electric fields, similar approaches may be used to enhance, inhibit or steer blood vessel outgrowth during angiogenesis. Chang et al.  have reported that cells from corneal endothelium – the epithelial cell layer on the deep surface of the cornea – respond to electrical fields, but the effects on true vascular endothelia are unknown.
In the present study, we demonstrate that endothelial cells do indeed respond to electric fields, migrating specifically toward the cathode. In order to understand the underlying molecular mechanism for this effect, we also examine the reorganization of the actin cytoskeleton during field-induced migration. Endothelial cells normally migrate by projecting broad, actin-filled lamellipodia ; so one way an electric field could steer endothelial migration is by polarizing the assembly and/or movement of actin filaments within the cell. We show specific changes in the amount and distribution of actin filaments in electrically stimulated cells that indicate that electric fields can re-orient endothelial cells by redistributing sites of actin filament assembly.
Materials and Methods
Bovine aortic endothelial cells (BAECs) were obtained by trypsinizing the lumen of the aorta of a freshly slaughtered cow as described by Gajdusek and Schwartz . The cells used in this study were descendants from a single clone, which was confirmed as endothelial by positive staining for von Willebrand factor, factor VIII, and LDL receptor. The culture medium consisted of Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, and all experiments were performed on cells that were between passages 12 and 18 from the original isolation.
In order to monitor the behavior of cells during the application of an electric field, the cells were allowed to attach to 22 × 22 mm glass coverslips and mounted in a special chamber for microscopic observation (fig. 1). The chamber fit on the stage of a Zeiss Axiovert 135 inverted microscope was equipped with a long working-distance condenser. Precise electric fields were generated by connecting the chamber electrodes to a conventional power supply for gel electrophoresis (Pharmacia Biotech, Piscataway, N.J., USA), and the voltage drop across the cell field was measured with a voltmeter. A similar device consisting of four chambers in parallel, each with a longer channel to accommodate multiple cover glasses, was used when preparing multiple samples for fixation and staining.
Fig. 1. Chamber for application of electric fields. Electric field experiments were conducted in a Plexiglas chamber consisting of two large reservoirs connected by a narrow channel with platinum wire electrodes located at opposite ends of the reservoirs. The reservoirs were filled with a gel formed by dissolving 2% agar in the cell culture medium. The agar gel and long reservoirs prevented any electrolysis products produced at the electrodes from reaching the cells over the elapsed time of our experiments. A cover glass with attached cells was mounted over the connecting channel after filling the channel with culture medium. The small cross-sectional area of the connecting channel permitted a large voltage drop across the cells while minimizing the voltage required across the entire chamber and the amount of current that must be passed through the chamber. A glass window opposite the cover glass permitted transillumination of the connecting channel, and the entire chamber fit on the stage of an inverted microscope so that cell behavior could be observed directly during application of the electric field.
For microscopic observations, the culture medium in the cell chamber was equilibrated with 5% CO2 immediately before assembly, and the chamber was sealed with vacuum grease to prevent evaporation and to maintain carbonate buffering of the medium at pH 7.4 ± 0.2 (as determined from phenol red indicator dye in the medium). An air curtain was used to maintain the temperature at the center of the chamber at 38°C, with the heater controlled by a thermistor on the cell chamber. The microscope objective was also warmed (using an electric heating tape) to help maintain chamber temperature when using oil-immersion objectives. Temperature was monitored using a micro-thermistor (Physitemp Instruments, Clifton, N.J., USA) inserted directly into the chamber or placed in the immersion oil immediately outside the field of observation, and temperature fluctuations did not exceed ±1°C in these experiments.
Microscopy was performed on a Zeiss Axiovert 135 microscope using ×40 and ×100 Plan-Neofluar oil-immersion objectives or ×20 dry objectives with further ×1.6 magnification between the objective and camera. For time-lapsed imaging of live cells, the interval between images ranged from 30 s to 5 min, depending on the number of fields under observation and the length of the experiment. Transmitted light images were recorded digitally by integrating 32 frames of the video output from a DAGE-MTI video camera at a spatial resolution of ≤0.45 μm/pixel using a 16-bit Perceptics Pixelpipe acquisition board. Fluorescence images were recorded similarly from the video output of an intensified CCD camera (Paultek Instruments). All measurements were performed using Oncor-Image software on Macintosh desktop computers.
The rate of cell migration was measured from the displacement of the nucleus of the cell over time. We used the displacement of the nucleus, rather than the centroid of the cell, because it was less sensitive to fluctuations in cell shape. The position of the nucleus did not change when a cell extended and retracted protrusions without moving, but it did move when protrusion was accompanied by retraction of an opposite edge resulting in actual translocation of the whole cell. Speeds reported were calculated from the net displacement of the nucleus over a 15- or 30-min interval, depending on the duration of the experiment. Because the cells used in this study moved slowly and changed direction infrequently, net speeds calculated over these intervals were not significantly different from speeds calculated using paths traced with higher temporal resolution (down to as little as 30 s between images).
The distribution of lamellipodia was represented as a vector using a modification of the method described by Kolega . A simplified example of this calculation is illustrated in figure 2. Briefly, the image of a cell was superimposed on a grid of lines oriented parallel to the image frame. Each intersection between a horizontal grid line and the outline of the cell was scored for the presence of a lamellipodium extending toward the right (+1) or the left (–1). A region was considered a lamellipodium if it was flattened (as judged by phase microscopy), excluded membrane-bound organelles, and had a convex contour. The scores were summed and multiplied by the grid spacing to obtain the horizontal component of the vector. Vertical grid lines were scored similarly to obtain the vertical component of the vector. The vector sum of the horizontal and vertical components points in the direction of the net asymmetry of the distribution of lamellipodia around the cell. In addition, the magnitude of the vector indicates the length of the cell periphery that is extending in this direction ‘unopposed’, i.e. with no equivalent extension on the opposite side of the cell.
Fig. 2. Vector representation of the distribution of lamellipodia. a A grid of horizontal lines was superimposed on a micrograph of a cell. In this example, only two lines are shown, but in all experimental measurements 20–30 parallel, uniformly spaced lines were used so that the spacing between lines was approximately 1 μm. Arrowheads mark all the intersections between the grid lines and the outline of the cell. Each intersection was scored as follows: if there was a lamellipodium present with free space toward the right, the point was scored as +1; if there was a lamellipodium with free space to the left, the score was –1; and if no lamellipodium was present, the score was 0. b A set of vertical lines was scored similarly, with +1 indicating a lamellipodium facing the top of the image, –1 toward the bottom. c The horizontal and vertical scores were summed to give the horizontal and vertical components (dotted arrows) of a vector (solid arrow) that represents the net asymmetry in the distribution of lamellipodia around the perimeter of the cell. Note that this vector points in the ‘average direction’ of the lamellipodia, which extend mostly from the edges of the cell that face the top and right of the image.
F-actin content was measured by fluorescence microscopy. Cells on cover glasses were fixed for 30 min with 3.7% formaldehyde in a buffer consisting of 127 mM NaCl, 5 mM KCl, 1.1 mM Na2PO4, 0.4 mM KH2PO4, 4 mM NaHCO3, 2 mM MgCl2, 2 mM EGTA, 5 mM PIPES, 0.1% D-glucose, pH 6.0. In all experiments, the time that elapsed between turning off the electric field and immersing the coverslip in fixative was less than 60 s. After fixation, the cells were permeabilized with 0.05% Triton X-100 in the same buffer, then stained with 1 U/ml Oregon-Green phalloidin (Molecular Probes, Eugene, Oreg., USA) to visualize F-actin, and 1.0 μg/ml Texas red to label total cell protein . Fluorescence images of each fluorophore were acquired, and single cells in each field were traced manually using Oncor-Image software. A measure of F-actin content was obtained by summing the total intensity in the outlined region on the Oregon-Green image. This measurement was corrected for background fluorescence by subtracting the signal from an identically sized region in a field containing no cells. The total protein was measured in identical fashion from the Texas red image of the same cell. The ‘relative F-actin’ content was then obtained by dividing F-actin by total protein, thereby compensating for differences in cell size. Because these measurements were acutely sensitive to small variations in staining and imaging conditions, whenever different groups of cells were to be compared, they were all stained together with the same solutions and were imaged in the same session with identical illumination and camera settings. Illumination and camera settings were adjusted so that the maximum fluorescence intensities fell within the linear dynamic range of the imaging system.
F-actin asymmetry was measured as described by Coates et al. , except that the center of mass of F-actin was compared to the center of mass of total protein instead of the geometric center of the projection of the cell (fig. 3). This measurement reflects local enrichment of F-actin within the cell, as opposed to bulk movement of cytoplasm.
Fig. 3. Measurement of F-actin asymmetry. a Texas red staining of total protein distribution in a BAEC. Staining is most intense in the thick central part of the cell; i.e. in the nucleus and perinuclear cytoplasm. The calculated center of the fluorescence intensity is marked by the ‘o’, and is located close to the geometric center of the cell. b The same cell was also stained with Oregon-Green phalloidin to show F-actin. F-actin is particularly abundant in the flattened protrusion that extends toward the lower right. The calculated center of the fluorescence intensity is indicated by the ‘x’. c The centers of total protein (o) and F-actin (x) are superimposed on an outline of the cell. Note that the F-actin distribution is shifted toward the lower right. The size and direction of this shift (arrow) provides a vector measurement of the degree and orientation of the asymmetry in the distribution of F-actin in the cell.
We examined BAEC migration 4–8 h after the cells were trypsinized and seeded on glass coverslips. This interval is well after the cells have attached and spread, but before they form extensive extracellular matrix and cell-substratum adhesions that decrease their rate of translocation. Time-lapse imaging showed that even during this period, the cells migrated relatively slowly with an average speed of only 0.27 ± 0.03 μm/min (n = 55). In the absence of an externally applied electric field, cells migrated with equal probability in all directions, as demonstrated by the orientations of the paths traversed by randomly selected cells over a 15-min interval (fig. 4a).
Fig. 4. Oriented migration of BAECs in electric fields. The movements of single BAECs were followed by time-lapse video microscopy for 30 min beginning 6 h after the cells were seeded on cover glasses. Cells were seeded at low density and only cells that remained at least 10 μm from any other cell were tracked. In each panel, arrows show the distance and direction migrated by 20–30 cells over a 15-min interval. a Net migration over 15 min of observation, during which no electric field was applied. There was no preferred orientation to the movements of the cells. b Net migration by the same 25 cells shown in a over the next 15 min, during which a field of 10 V/cm was continuously applied. Note that the migration of every cell was skewed toward the cathode, even though the cells migrated randomly during the previous 15 min. c–e Net migration by cells over the second 15-min interval (i.e. with electric field on) in similar experiments with fields of 1, 2, and 5 V/cm, respectively. f The CMI, (circles, solid line) and speed of migration (triangles, dashed line) were calculated for the conditions in a–e. The CMI equals the distance traveled in the cathodal direction divided by the total distance traveled; speed of migration was measured from the displacement of the centroid of the cell nucleus. Each point represents the average of 20–30 different cells, and error bars indicate a range of one standard error. * p < 0.01, significantly different from 0, ** p < 0.01 speeds that are significantly greater than controls (0 V/cm). A separate cell preparation was used at each field strength, so speeds are expressed as a percentage of the speed of the same cells measured during the 15 min immediately prior to the application of the electric field. Note that a more than tenfold increase in the CMI occurs between 1 and 2 V/cm, whereas the speed of migration was not affected by fields of up to 5 V/cm. At 10V/cm, cells did move ∼60% faster than cells in weaker fields.
In contrast, when cells were exposed to a 10 V/cm electric field, they all migrated towards the cathode (fig. 4b). Electric fields ≤1 V/cm did not produce this effect on cell migration (fig. 4c), but most cells did orient when the field was increased to 2 V/cm, and all cells were oriented at 5 V/cm. This response is quantified in figure 4f using a measure of orientation that we called the cathodal migration index, or CMI. The CMI is the ratio between the displacement of the cell towards the cathode and the total displacement of the cell during the same time interval, with the displacements determined from the position of the centroid of the nucleus at the beginning and end of a 15-min treatment interval. For cells migrating in the absence of an external electric field or at 1 V/cm, the CMI was not significantly different from 0 (0.09 ± 0.14 and 0.08 ± 0.12, respectively; p > 0.5), indicating no preferred orientation. However, at 2 V/cm or greater, the CMI was significantly positive (0.57 ± 0.10, n = 21; p < 0.01), indicating that migration was biased toward the cathode. When the average direction of the cell displacements is compared with the orientation of the electric field using the statistical test of Stephens  for comparing directions, it is not significantly different from the direction of the cathode (p < 0.01) at 2, 5, and 10 V/cm.
The development of a preferred orientation was not accompanied by an increase in the speed of migration. Even though migration was strongly oriented toward the cathode at 2 and 5 V/cm (fig. 4f, solid line), the absolute rate of migration did not change significantly until the field strength reached 10 V/cm (fig. 4f, dashed line). Thus, at a threshold between 1 and 2 V/cm, electric fields affected the steering mechanism of the cell without stimulating overall motor activity.
Since BAECs normally move over flat surfaces by extending lamellipodia from their leading edges, we used time-lapse video microscopy to determine if electric fields selectively stimulated lamellipodia facing the cathode and/or suppressed lamellipodia extending in other directions. Unstimulated BAECs typically had multiple lamellipodia, which were small and irregularly distributed around the cell perimeter (fig. 5a–c). These lamellipodia extended, retracted, and ruffled, without any apparent coordination, producing sporadic displacements of the cell with the random orientations seen in figure 4a.
Fig. 5. Redistribution of lamellipodia in an applied electric field. BAECs were allowed to attach to glass coverslips for 6 h , then were observed by differential interference contrast microscopy during the application of a 5 V/cm electric field. a–c A single BAEC immediately before the field was turned on. Numerous lamellipodia (small arrows) were present, but their movements were limited to minor (<2 μm) extensions, retractions and ruffles. Note that the lamellipodia faced in several different directions, so that there is little overall asymmetry in the protrusive activity of the cell. d–f The same cell after a 5 V/cm field (in the direction of the small arrow) was turned on. Within 15 min, existing lamellipodia on the side of the cell facing the cathode enlarged and merged (large arrows), and new lamellipodia extended from previously inactive regions (arrowheads). Lamellipodia were completely withdrawn from the anodal side of the cell. By 30 min (e), the cell had migrated ∼10 μm toward the cathode. Large lamellipodia persisted on the entire cathodal edge of the cell, but were still absent from the anodal edge.
When fields ≥2 V/cm were applied, two events occurred simultaneously: new lamellipodia formed along the edge of the cell closest to the cathode, and existing lamellipodia were retracted along the edge of the cell facing away from the cathode (fig. 5d–e). New lamellipodia often sprouted from edges that previously had no protrusive activity (fig. 5d–e, arrowheads), and these ‘fresh’ lamellipodia could be detected as early as 2 min after the field was turned on. In addition, existing lamellipodia that already faced the cathode extended further towards the cathode and increased in size while the field was on. Once lamellipodia were oriented by an electric field, the asymmetric distribution tended to persist immediately after the field was turned off, with random orientation gradually returning over a 30- to 60-min period. However, if the polarity of the electric field was reversed, lamellipodia began to rearrange in less than 2 min. Lamellipodia facing the new anode began to withdraw and new lamellipodia began to appear along cathode-facing edges, with complete reversal of lamellipodial polarity occurring within 15 min at 10 V/cm.
The effect of electric fields on the distribution of lamellipodia was quantified by measuring the net orientation of lamellipodial protrusions around the cell periphery as described in the Materials and Methods. Fields that caused cells to migrate toward the cathode (i.e. fields ≥2 V/cm) also caused lamellipodia to become skewed toward the cathode (fig. 6). In fields ≥2 V/cm, the average of the vectors representing lamellipodial distribution pointed in a direction that was not significantly different from the cathode by the test of Stephens  (p > 0.99). This is consistent with the well-established role of lamellipodia as the primary locomotive organelle when endothelial cells migrate over flat surfaces .
Fig. 6. Orientation of protrusive activity at different electric field strengths. Microscopic images of single BAECs were acquired at 5-min intervals while the cells were exposed to an electric field for 30 min, beginning 6 h after seeding. The net distribution of lamellipodial protrusions at the end of the 30-min period was then measured as described in the Materials and Methods. a–e Vector representations of the net protrusive activity of individual cells exposed to 0, 1, 2, 5, and 10 V/cm, respectively. Each arrow corresponds to one cell, and it points in the direction of the net asymmetry in the orientation of lamellipodia on that cell. In the absence of an applied field (a) and at 1 V/cm (b), lamellipodia had no preferential orientation. In contrast, lamellipodia were skewed toward the cathode in fields ≥2 V/cm (c–e). f Dependence of protrusive asymmetry on electric field strength. The magnitude of the average vector was calculated at each field strength and broken down into the component that was oriented towards the cathode (circles, solid line) and the perpendicular component (triangles, dashed line). Error bars represent one standard error from the mean, and asterisks indicate values that are significantly different (p < 0.01) from 0. Note that protrusion toward the cathode rises sharply between 1 and 2 V/cm, whereas the average protrusive activity perpendicular to the field does not deviate significantly from 0.
In normal migrating cells, lamellipodia extend because actin filaments assemble just inside the advancing edge of the protrusion. Treatment of BAECs with cytochalasin D to prevent actin assembly inhibited the formation of lamellipodia and blocked protrusion in response to an electric field (fig. 7). Thus, electric fields may direct lamellipodial protrusion by inducing de novo assembly of actin filaments in the side of the cell facing the cathode. To test this possibility, the distribution of filamentous actin (F-actin) during field-induced migration was examined by fluorescence microscopy. Figure 8 illustrates the typical F-actin and total protein distributions in an unstimulated cell and in a cell exposed to 5 V/cm for 30 min. Unstimulated cells contained many stress fibers that formed a complicated array of radially and circumferentially oriented bundles (fig. 8a). F-actin was also present between the stress fibers as a diffuse network that was not resolved as discrete structures in the light microscope. This diffuse staining was particularly apparent in lamellipodia (fig. 8a, arrowheads), which were much flatter than the cell body and contained less protein mass (compare fig. 8b), but nonetheless stained as strongly for F-actin as the thicker perinuclear cytoplasm. Cells migrating in a 5 V/cm field also contained an abundance of stress fibers and diffuse staining between fibers (fig. 8c). The size and number of stress fibers did not differ dramatically from unstimulated cells, but fibers appeared to be more abundant in the half of the cell that was closer to the cathode. In addition, the staining between fibers in the cathodal cytoplasm appeared to be more intense, especially in lamellipodia at the leading edge of the cell.
Fig. 7. Inhibition of electric-field-induced protrusion by cytochalasin D. BAECs were mounted for observation in culture medium containing 1 μM cytochalasin D observed for 60 min in the continued presence of drug beginning 15 min after addition of the drug. a–c During the first 30 min of observation, no electric field was applied. By the time the first image (a, t = 0 min) was acquired, cytochalasin D has suppressed protrusive activity and most existing lamellipodia were already retracted, leaving the cell with many narrow, highly branched extensions. Over the next 30 min, the cell did not extend any new lamellipodia or filopodia in any direction. d–f A 5 V/cm electric field was applied in the direction indicated by the arrows. The field had little or no effect on protrusive activity: there were no detectable lamellipodia or filopodia extended toward the cathode, nor was there any significant withdrawal of cytoplasm away from the anode.
Fig. 8. F-actin distribution in BAECs responding to a 5 V/cm electric field. BAECs were allowed to attach to cover glasses for 6 h, then placed in a chamber for 30 min with no applied field (a, b) or in an electric field of 5 V/cm (c, d). Cells were immediately fixed and stained with Oregon Green phalloidin to visualize F-actin (a, c) and Texas red to show total protein (b, d). a, b Unstimulated cells had numerous fine actin bundles as well as regions of diffuse F-actin staining that indicated filament networks that were not resolved at the light-microscopic level. F-actin bundles had no predominant orientation, and actin-rich lamellipodia extended in multiple directions (arrowheads). c, d During 30 min in a 5 V/cm field, a cell developed a broad, flat protrusion (brackets) on the side of the cell facing the cathode (arrow at lower right indicates the orientation of the electric field). The protrusion contained high levels of both F-actin bundles and diffuse F-actin staining, whereas the perinuclear cytoplasm contained relatively little F-actin other than large bundles along the cell edges (arrowhead). Note the opposite distribution of total protein (d); i.e. staining was greatest in the nucleus and the surrounding cytoplasm where the cell is thickest, and less intense in the flat protrusion at the front.
Visual inspection of many such images suggested that the electric field caused an increase in actin filament assembly, and that assembled actin was preferentially located toward the negative pole of the field. To verify that the electric field did indeed stimulate actin assembly, the amount of F-actin relative to total protein was measured by quantitative fluorescence imaging of BAECs exposed to a 5 V/cm field for various times. Within 15 min of applying the field, F-actin content increased by 50% compared to unstimulated cells, and it had nearly doubled after 30 min (fig. 9). However, this effect was transient. If cells were maintained in a 5 V/cm field for 60 min, the relative F-actin concentrations returned to unstimulated levels. This was not due to any deleterious effect of the field or to an accommodation of the cells to its continued application, as cells continued to move toward the cathode for at least 120 min with no loss of speed or polarity.
Fig. 9. F-actin assembly in BAECs responding to a 5 V/cm electric field. An electric field of 5 V/cm was applied to BAECs beginning 6 h after seeding. At selected intervals, cover glasses were fixed and stained for determination of the relative F-actin concentration as described in the Materials and Methods. Each point on the graph represents the mean value from three cover glasses, with 25–30 randomly selected cells measured on each cover glass. Error bars indicate one standard error from the mean. F-actin levels that are significantly greater than controls are marked (* p < 0.01).
If the electric field caused actin to assemble specifically in the cathode-facing side of the cell – and did not simply shift bulk cytoplasm toward the cathode – then the distribution of F-actin should be preferentially skewed toward the cathode when compared to other protein in the cell. This was tested by comparing the center of mass of F-actin to the center of mass of total protein, as described in the Materials and Methods. Even in unstimulated cells, the center of mass of F-actin did not correspond to the center of mass of total protein: the absolute distance between the two centers averaged 5.4 ± 0.5 μm (n = 29). This indicated that F-actin was asymmetrically distributed even in unstimulated cells, and presumably reflects the local enrichment of F-actin in lamellipodia as cells migrate randomly. More importantly, this asymmetry had no preferred orientation, as shown by the distribution of the F-actin centers of mass, which lay in all directions around the total protein center (fig. 10a). The average net displacement of the F-actin center away from the protein center was not significantly different from 0 (p > 0.5) in either the x direction (–0.5 ± 0.8 μm) or the y direction (0.6 ± 0.7 μm), confirming the absence of a preferred orientation in the F-actin asymmetry. In contrast, when cells were exposed to 5 V/cm for 30 min, the distribution of F-actin became strongly skewed toward the cathode (fig. 10b). Both the average absolute magnitude of the asymmetry (i.e. the average distance between the F-actin center and the protein center) and the average net displacement specifically toward the cathode were significantly greater than in unstimulated cells (9.8 ± 1.3 and 6.4 ±1.4 μm, respectively, n = 29; p < 0.01). There was no change in asymmetry perpendicular to the electric field; net displacement in this axis was not significantly different from 0 (0.5 ± 1.3 μm). Thus, the average position of the F-actin center was 6.4 μm away from the total protein center along a line that was just 4.5° away from the cathodal direction. Using the statistical test of Stephens , the direction of this average displacement and the actual cathodal direction are not significantly different (p > 0.99).
Fig. 10. F-actin asymmetry in BAECs responding to a 5 V/cm electric field. BAECs were allowed to attach to cover glasses for 6 h, then placed in a chamber with no applied field for 30 min (a), in an electric field of 5 V/cm for 30 min (b), or in an electric field of 5 V/cm for 2 h (c). Cells were immediately fixed and stained for total protein and F-actin. The asymmetry in the distribution of F-actin relative to total protein was then calculated for individual cells as described in figure 3. Each arrow represents the location of the center of mass of F-actin for a single cell, with the origin (0, 0) corresponding to the center of total protein. Note the random orientation of F-actin asymmetry in unstimulated cells (a), and the strong bias toward the cathode in the presence of a 5 V/cm electric field (b, c). The effects are summarized in d: the absolute asymmetry is the average distance between the F-actin center and total protein center; the net asymmetry is the component of this displacement in the direction of the cathode; and the cathodal asymmetry index is the displacement toward the cathode divided by the total displacement. Each bar represents the mean measurement from 25–30 cells, with error bars representing one standard error. All values at 30 and 120 min are significantly greater than the corresponding measurement at 0 min (p < 0.01).
The degree of asymmetry was quantified with a ‘cathodal asymmetry index’, analogous to the CMI used to assess the directionality of movement. The cathodal asymmetry index was defined as the net asymmetry of the F-actin center toward the cathode (i.e. the component of the displacement between the F-actin center and the protein center that lies parallel to the electric field) divided by the absolute asymmetry (i.e. the whole distance between the F-actin center and the protein center). These measurements of F-actin asymmetry are summarized in figure 10d. All three measures of F-actin asymmetry – the absolute asymmetry, the net bias toward the cathode, and the cathodal asymmetry index – were significantly different (p < 0.01) from control cells after 30 min of exposure to a 5 V/cm electric field.
Unlike the amount of F-actin, which returned to basal levels after 60 min even in the continued presence of a 5 V/cm electric field, the asymmetric distribution of the F-actin persisted (fig. 10d). After 120 min at 5 V/cm, the average absolute F-actin asymmetry was even larger than at 30 min, increasing to 14.1 ± 1.4 μm (n = 28). The net asymmetry towards the cathode was also larger (12.2 ± 1.6 μm), but the net asymmetry in the direction perpendicular to the field was again not significantly different from 0 (0.1 ± 0.9 μm; p > 0.9). This corresponds to an average asymmetry that is oriented only 0.5° from the axis of the electric field, which is not significantly different from the direction of the cathode (p > 0.99).
Many different kinds of cells, ranging from amoebae to neurons, have been shown to orient in electric fields , but the direction of field-induced movement is dependent on the cell type. Cells that migrate toward the cathode include corneal  and epidermal [21, 22] keratinocytes, pigmented retinal epithelium , neural crest cells , embryonic fibroblasts , and osteoblasts , whereas corneal endothelium  and corneal fibroblasts , peritoneal macrophages , and osteoclasts  all move toward the anode. Our current results show that BAECs migrate toward the cathode, and we have seen similar responses from microvascular endothelial cells derived from bovine adrenal medulla and venous endothelial cells from human umbilicus [unpubl. observations].
The electric field around a wound in skin or cornea has its cathode toward the center of the wound [5, 6]. Therefore, endothelial cells at the periphery of the wound would be directed to migrate into the wound, promoting angiogenesis to restore the vasculature of the damaged tissue. Barker et al.  have measured the field in skin wounds and found it to exceed 1 V/cm. We show that the threshold at which endothelial cells begin to migrate in response to an external field in vitro is 1–2 V/cm (fig. 4). Thus, physiological electric fields may contribute to the regulation and steering of endothelial migration, supplementing or modulating the effects of the well-known chemical signals that promote angiogenesis. Interestingly, solid tumors have also been shown to generate electric fields with the tumor more negative than the surrounding tissue . The magnitude of these field-associated changes is not precisely known, but the direction is such that endothelial cells would be induced to move toward the lesion as in surface wounds. Thus, electric fields may also play a role in stimulating tumor angiogenesis.
Persistent, directed migration of endothelial cells requires polarized protrusion and retraction. Our time-lapse observations show that electric fields affect both: protrusions facing the cathode are promoted, whereas anode-facing protrusions are suppressed (causing retraction). The different effects on the front and rear of the cell produce and maintain the asymmetric distribution of protrusions that is characteristic of highly directional migration. A similar response has been described for chick embryo fibroblasts  and Xenopus epidermal cells . Fish epidermal cells exhibit a more gradual turning , but the end result is the same: preferential distribution of lamellipodia toward the cathode and retraction of surfaces facing the anode. The rate at which lamellipodia extend and retract is apparently not affected by the electric field. Even at 5 V/cm, which is at least twice the threshold required to orient movement, BAECs did not move significantly faster than unstimulated cells. Thus, the primary effect of the electric field is on the location of the locomotive machinery, not how fast it operates. We did observe an increase in the speed of migration at high field strength (10 V/cm), which probably reflects additional morphological changes that occur when lamellipodial distribution is completely asymmetric. That is, when all lamellipodia are withdrawn from the anode-facing side of a cell, the cell tends to be stretched perpendicular to the field by the protrusions at the ends of the affected area. This makes the cathode-facing surface much broader too, and the cell develops lamellipodia that extend over a region that is much wider than a normally migrating cell. This exaggerated leading edge may permit the cells to move faster because more of the bulk cytoplasm could be moving forward at any given time.
Cathodal protrusion and anodal contraction occur prior to changes in the gross morphology of the intervening body of the cell, suggesting that the electric field acts first at the periphery of the cell. The main component of peripheral cytoplasm is the actin cytoskeleton, and it is generally accepted that changes in actin organization are responsible for the extension and retraction of lamellipodia and other locomotive protrusions during normal cell crawling. Electric-field-induced protrusions are indistinguishable from normal cell protrusions, and a wide band of actin filaments is found at or near their leading edges in fibroblasts [22, 25], retinal and corneal epithelial cells [22, 29], and fish keratinocytes , as well as the vascular endothelial cells in the present study. Thus, a key element in understanding how an electric field polarizes cell migration is understanding its effect on actin assembly.
Zhao et al.  used confocal imaging to demonstrate in galvanotaxing corneal epithelial cells that the density of actin filaments is higher at the front than in the rear. By comparing actin filament distribution with protein mass, we show that this is due to a selective accumulation of actin filaments in the cathodal cytoplasm, and not to a general concentration of cytoplasmic protein in the direction of migration. Furthermore, we show that electrical stimulation caused a large increase in the total amount of filamentous actin in the cell, indicating that reorientation in an electric field involves the assembly of new actin structures. Preferential localization of these new structures in the cathodal cytoplasm would explain the field-induced shift in F-actin distribution. Consequently, we believe that a crucial signaling event between the galvanotactic electric field and the locomotive machinery of the cell is localized stimulation of actin assembly.
A number of different signaling pathways that could elicit changes in actin assembly have been implicated in the response of cells to electric fields. One of the foremost potential signals is Ca2+. A rise in intracellular free Ca2+ has been observed when fibroblasts  and neuroblastomas  are exposed to galvanotactic electric fields, and preventing Ca2+ influx or intracellular Ca2+ release can block field-induced migration and shape changes in keratinocytes and fibroblasts [21, 31]. An electric field could produce local differences in the concentration of free Ca2+ because it depolarizes the plasma membrane facing the cathode and hyperpolarizes the membrane facing the anode , which would cause differential opening of voltage-sensitive Ca2+ channels on opposite sides of the cell . Elevated free Ca2+ in the cathodal cytoplasm could then turn on the actin-severing activity of gelsolin, thereby creating more actin-filament ends for assembly. Ca2+ also activates protein kinase C, and activation of protein kinase C stimulates cell spreading and peripheral actin assembly in a variety of systems . Protein kinase C phosphorylates several cytoskeletal targets, but the precise mechanism by which it induces assembly and extension is unknown. However, inhibitors of protein kinase C do prevent neural crest cells from orienting in an electric field . These inhibitors have no effect on continued directed migration if administered after the cells have already polarized, suggesting that protein kinase C is required for rearrangement of the locomotive machinery, but is not necessary for the continued activity of the machinery.
A transient, protein-kinase-C-mediated rearrangement of the machinery is consistent with our observation that F-actin levels rise as cells re-orient, then returned to baseline in the continued presence of an electric field. The fact that migration toward the cathode persisted after F-actin assembly declined indicates that galvanotaxis does not require continuously elevated actin assembly. The transient increase in F-actin may reflect temporary recruitment of the free actin of the cell into new protrusions sprouting toward the cathode, before significant portions of the existing actin cytoskeleton can be disassembled. The subsequent decrease may reflect accommodation of the cell to the electric field or changes in gene expression induced by the field. However, the rapid reversal of protrusive activity that occurs if the field is reversed suggests that such longer-term processes do not effect the ability of the cell to re-orient and are not essential to the process. In any case, once cells are fully polarized and the entire cell begins to move consistently toward the cathode, a new equilibrium is established between assembly at the front and disassembly at the rear. Because actin filaments in protrusions are oriented with their plus ends toward the leading edge and their minus ends toward the rear, the actin cytoskeleton would be expected to maintain the same general polarity even as the individual filaments continuously assemble and disassemble. The persistent asymmetry in the distribution of F-actin, even after the increase in actin assembly has subsided, is consistent with such a model. Directional migration may thus be self-sustaining once an overall cell polarity is established. In this regard, it is noteworthy that cathode-facing protrusions are not immediately withdrawn when the electric field is turned off, even though they are rapidly retracted if the cell is actively re-oriented by reversing the polarity of the field. Although we have not examined this phenomenon in detail in endothelial cells, Cooper and Schliwa  have shown that epidermal keratinocytes migrating in an electric field will continue to move towards the cathode for many cell lengths after the field has been shut off, only gradually returning to randomly oriented movement.
It must be noted that other laboratories have reported that fibroblasts , neurons  and prostatic carcinoma cells  can orient to an electric field without Ca2+ fluxes. Thus, there must be alternative ways to signal cytoskeletal reorganization. A major possibility is that cell surface components are electrophoresed in the plasma membrane, accumulating at (or being depleted from) the extreme edges of the cell . Receptors for concanavalin A [25, 35], fibronectin and epidermal growth factor  all redistribute toward the cathode during field-induced migration. Concanavalin A and fibronectin receptors have transmembrane linkages to the actin cytoskeleton, and the epidermal growth factor receptor is itself an actin-binding protein , so the redistribution of these components may directly move actin filaments and/or actin nucleation sites toward the cathode. In addition, clustering of integrins and growth factor receptors trigger their activation, switching on secondary pathways such as protein kinase C or protein kinase A. Protein kinase A has been implicated as a possible signaling mechanism by the observation that galvanotaxis of epidermal keratinocytes is not blocked by downregulating protein kinase C, but is sensitive to protein kinase A inhibitors . Protein kinase A indirectly activates rac , which induces actin assembly and the formation of lamellipodia in a variety of migrating cells .
The present study does not allow us to distinguish among these signaling mechanisms, but our observations indicate that the signal ultimately elicits spatially restricted assembly of actin filaments. Further examination of the specific local event(s) that initiate actin assembly should reveal more of the intermediate signaling steps. Of particular interest is whether new sites for actin polymerization arise via de novo nucleation, severing of existing filaments, or filament branching, and whether new assembly is confined to the cortical cytoplasm or occurs deeper in the cell as well. Answers to these questions will further define the endpoints of the signaling pathways that are essential for field-directed migration and will elucidate the steering mechanism of cell migration in general. Ultimately, manipulation of biological electric fields and/or the migratory response of cells to such fields may provide useful tools for precisely controlling cell movements in situ.
The authors thank Drs. C. Cohan, R. Hard, and F. Mendel for advice, support, and criticism throughout the progress of this work, and S. Kumar and M. Pazik for excellent technical assistance. Portions of this study were included in a thesis submitted by X. L. to the Graduate School of the University of Buffalo in partial fulfillment of the requirements for a Masters degree in Anatomy and Cell Biology.
Dr. John Kolega
Department of Pathology and Anatomical Sciences
University of Buffalo School of Medicine and Biomedical Sciences
3435 Main Street, Buffalo, NY 14214 (USA)
Tel. +1 716 829 3527, Fax +1 716 829 2911, E-Mail firstname.lastname@example.org
This work was supported by grants from the American Heart Association (0050232N) and the National Science Foundation (MCB9417115).
Received: Received: November 20, 2001
Accepted after revision: February 28, 2002
Number of Print Pages : 14
Number of Figures : 10, Number of Tables : 0, Number of References : 42
Journal of Vascular Research
Founded 1964 as Angiologica by M. Comèl and L. Laszt (1964–1973) continued as Blood Vessels by J.A. Bevan (1974–1991)
Vol. 39, No. 5, Year 2002 (Cover Date: September-October 2002)
Journal Editor: M.J. Mulvany, Aarhus
ISSN: 1018–1172 (print), 1423–0135 (Online)
For additional information: http://www.karger.com/journals/jvr