Macrophages sense and follow physiologically relevant K⁺ and Na⁺ gradients: how induced dipoles and ionotaxis may explain electrotaxis

Abstract

Background

Many cells respond to changes in concentration of extracellular molecules (chemotaxis) by directing migration towards increased concentration of molecules such as food or signaling molecules, or away from toxins. Similarly, some cells respond to extracellular electric fields by migrating towards cathode or anode (electrotaxis). However, only motile bacteria are known to respond to extracellular ion gradients (ionotaxis), with no similar reports in mammalian cells.

Methods

We used microfluidic gradient chips to examine the migration of human macrophages in concentration gradients of sodium, potassium and chloride.

Results

All three ion gradients significantly directed motion in the direction of the gradient compared to cell migration at constant concentration. Macrophages showed significant migration towards higher concentrations of cations. Gradients as low as ± 0.025% over ambient ion concentration could be sensed. The results suggest that macrophages use small changes in extracellular K + and Na+, to identify injury sites, given that increases in local K⁺ in the body would signify cell rupture.

Discussion

We also considered the implications of this effect for the phenomenon of electrotaxis. Electric fields induce dipoles at the cell surface, aligned with the field. If these interact with extracellular ions, they will generate local ion gradients across the cell, which the cell can then follow. Hence it is the induced gradient, not the electric field, which the cell interacts with.

Conclusions

This work suggests that local variation in ion concentrations may be a previously unexplored mechanism for cellular communication with ramifications for cell function, developmental biology and cancer.

Introduction

Many cells respond to extracellular stimuli by moving, growing, differentiating or dividing [1]. The orientation of cell movement in response to an extracellular chemical gradient is called chemotaxis [2–5], which has been studied widely due to its role in guiding immune cell migration, as well as in developmental biology [6, 7] and response to wounding [8–10]. Cellular responses to other extracellular stimuli have been reported, but are less well studied. For example, the alteration of behaviour by an ion gradient (ionotaxis) has been observed in bacteria [11], where bacterial colonies use K⁺ gradients to attract other bacteria in the vicinity [12], and also to communicate within the colony [13]. However, ionotaxis has not been reported in eukaryotes; the only report of the influence of ion gradients on mammalian cell behaviour described how neurons grown in interconnected chambers containing 5mM and 20mM K⁺ zones could grow neural processes in all directions at either concentration, but were unable to grow up the K⁺ gradient from the low-K⁺ to high-K⁺ chamber [14]. These two studies suggest that both neurons and bacteria respond mechanistically to ion (and specifically K⁺) gradients.

We hypothesised that macrophages are likely to respond to extracellular K⁺ gradients, since their function is to migrate to sites of injury; these are often associated with elevated K⁺ due to cell membrane rupture releasing high concentrations of intracellular K⁺ to the surroundings. [15]. Macrophages are also observed to migrate to tumours (becoming tumour-associated macrophages or MACs), which are also associated with high levels of extracellular K⁺ [16]. Macrophages express a range of ion channels which have been identified as regulating migration, polarisation and phagocytosis [17]; those implicated in macrophage migration include calcium-activated K⁺ channels [18, 19], voltage-sensitive K⁺ channels [18], the nonspecific cation channel TRP [20, 21], gap junction protein Connexin 43 [22], acid-sensing Na⁺ channels (ASIC) [23] and inwardly-rectified K⁺ channels [24].

To investigate this further, we examined the behaviour of macrophages in concentration gradients of K⁺, Na⁺ and Cl⁻, using commercial microfluidic chips to generate gradients in one ion species at a time. Migration was then observed to determine biases in migration, both in terms of movement along the gradient axis vs. orthogonal to it, and in terms of absolute direction up or down the gradient. We also used a model of dielectric properties of the cells to relate the behaviour of macrophages in ion gradients with that observed in electric fields [25]. By considering the relationship between imposed electric field, surface ion concentration and cell migration, our results strongly suggest that the phenomenon of electrotaxis is in fact a form of ionotaxis arising from an induced, localised ion gradient.

Ions are omnipresent throughout biology, and transmembrane ion gradients are the cornerstone of muscle and nerve function. Whilst ion gradients (largely in potassium) are a factor in injury, inflammation and cancer, little work has been performed on understanding how cells respond to local variations in ion concentration in their environment. Conversely, research on electrotaxis extends by over a century, yet no mechanism has been identified. Here, we use a simple model of the response of dielectric object to an imposed electric field. Analysis of this model suggests that cells in electric fields may induce, and subsequently follow, their own extracellular ion gradients. Given the long history of demonstrations of electrotaxis, this suggests ion gradients may explain many previously unexplored aspects of cell motility, developmental biology, and cancer.

Methods

Cell culture

Frozen THP-1 monocytes isolated from a one-year-old male were obtained from ECACC, UK and cultured in RPMI 1640 (Sigma Aldrich, UK) supplemented with 10% FBS (Sigma Aldrich, UK), 1% penicillin/ streptomycin (Sigma Aldrich, UK), and 2 mmol/L L-glutamine (Sigma Aldrich, UK). These were maintained in culture at a density up to 10⁶ cells/mL in vented flasks at 37˚C, 5% CO₂ and 95% humidity. Cells were passaged every three to five days, depending on cell density and condition. As cell population grew, cells were seeded in larger flasks up to T175 to accommodate.

Migration study

THP-1 monocytes were differentiated to macrophages ‘on chip’. The cell suspension was introduced to a Fluidic 834 gradient chip (Microfluidic Chip Shop, Jena, Germany) via the inlet port where the cells differentiated into adherent M0 type macrophages over a 72 h period. Subsequently, medium was refreshed to PMA free media and cells allowed to rest for 48 h prior to gradient introduction. During the differentiation process the chip was incubated at 37˚C, 5% CO₂ with inlet and outlet ports sealed to prevent bubble formation. For each biological repeat two chips were prepared, due to the prevalence of bubbles that would corrupt the internal migration area. Once M0 macrophages were adherent on the chip base, two solutions of differing ionic composition were introduced to the gradient chip via the inlet ports. The third inlet port was sealed with a mini Luer plug (Microfluidic Chip Shop, Jena, Germany) wrapped in parafilm. Solutions were loaded into 8 cm lengths of polyurethane tubing and attached to the two outermost inlet ports. Once solutions were loaded into the tubing, a 1 mL syringe with a 1 cm length of tubing was attached to the outlet port and the solutions were pulled through the chip, simultaneously removing culture media and creating an even gradient.

Cells were observed on the chip using a confocal microscope (Ti-Eclipse A1M confocal microscope, Nikon, Japan) using time-lapse imaging over a 14-hour period. Confocal time lapse imaging used a x10 objective, incubator set to 37˚C with a focal area scan of 6 × 8 and 15% overlap, an image was taken at 30-minute intervals. Analysis was carried out using the manual tracking plugin on the Image J software (NIH, Maryland, US). Two solutions from Table 1 were introduced in the outer inputs of the chip, forming a gradient across the main chamber. To compensate for non-gradient related phenomena such as flow, a control study of cell migration in an isotropic medium without a gradient was performed using complete culture medium. Three biological repeats were performed (four for K⁺ gradients) with 113 ± 56 cells analysed per experiment. In addition to tracking cells via timelapse, overall cell migration was also analysed by comparing the start and end position using ImageJ (NIH, Bethesda MD). A MATLAB (MathWorks, Natick NJ) script was used to convert this to a direction vector. Cells were then binned in quadrant (Fig. 1c) or bidirectional (Fig. 1d) splits.

Table 1.

The solutions used in this experiment. The solutions are arranged in pairs (marked as 1, 2, and 3), where each two solutions form the extremes of the ion gradient. In each case the medium marked B (B1, B2, B3) was at the bottom of the channel, the medium marked T (T1, T2, T3) was at the top of the channel

	Solutions
	(B1) 5 mM K+	(T1) 15 mM K+		(B2) 107 mM Na+			(T2) 127 mM Na+		(B3) 133 mM Cl−		(T3) 153 mM Cl−
Salts	Salt Concentration in DI water (mM/L)
KCl	5	15		5			5		5		5
NaCl	125	125		105			125		125		125
CaCl2	1.5	1.5		1.5			1.5		1.5		1.5
Na3PO4	0.8	0.8		0.8			0.8		0.8		0.8
[(CH3)3 NCH2CH2OH]+ +Cl−	10	0		30			10		0		20
C12H22O11	0	0		0			0		40		0

Ions	Ion concentration (mM/L)
K+	5	15		5			5		5		5
Na+	127.4	127.4		107.4			127.4		127.4		127.4
Ca2+	1.5	1.5		1.5			1.5		1.5		1.5
Cl−	143	143		143			143		133		153
PO43−	0.8	0.8		0.8			0.8		0.8		0.8
[(CH3)3 NCH2CH2OH]+	10	0		30			10		0		20
C12H22O11	0	0		0			0		40		0

Fig. 1.

Experimental setup. a A photograph of the experimental setup. Gradient chips (Microfluidic Chip Shop, Jena, Germany) were connected to polythene tube reservoirs, whilst 0.9 µL/min fluid flow was driven by capillary action due to evaporation from the outlet port. b A schematic of the gradient area (the circled zone in (a)) where measurements were made. Cells were observed microscopically in two illustrated zones to avoid artefacts. The main zone (dashed line) covering 50% of the chamber width and 50% of the length, whilst analysis of the central zone (dotted line) focussed on the central 18% of chamber width (50% of length as before). c and d Directions used in the study; positive x denotes the direction of flow, positive y denotes increasing ion gradient. Cell migration was grouped in either quadrants (c), or direction up/down the gradient (d). In each axis, the origin represents the stating position of each cell analysed

The main chamber (focus image in Fig. 1) had dimensions of 10 mm along the direction of flow, 8 mm across the flow, and 0.8 mm high. The combined flow rate of the two channels was approximately 0.9 µL/min. Flow was maintained by evaporation from the outlet. Macrophages contain shear-sensitive channels such as TRPV4 or Piezo1 that can be activated in flow at shear stresses above 1–2 dyne/cm². To address this, we calculated the shear stress on the macrophages in the experiment using established principles for laminar parallel flow in a rectangular duct [26]. Assuming a dynamic viscosity of 0.001 Pa s for the suspending medium yielded a peak shear stress in the chamber (mid-way on the side wall) of 0.15 dyne/cm², an order of magnitude below threshold.

DiSBAC23 live cell staining and imaging

Cells were aliquoted into a 24-well plate (Thermo Fisher Scientific, UK) at a density of 2.5 × 10⁵, 8 wells were allocated to each biological repeat. 100 nM of PMA (Sigma Aldrich, UK) in culture media was used to differentiate monocytes into M0-type macrophages over 60 h, 37˚C, 5% CO2, 95% humidity. After 60 h, cells were rested in PMA free culture media for a further 24 h prior to experimentation. The protocol for DiSBAC23 assessment was adapted from that reported by Li et al. [27]. Culture medium was removed and replaced with one of the solutions used during the migration studies. The formulation of the solutions is described in Table 1. 100 nM DiSBAC23 (excitation wavelength 530 nm, emission wavelength 560 nm) was added to each solution, and the cells incubated in these solutions at 37˚C for 30 min before confocal imaging. Three biological repeats were taken for each ionic solution. Images were taken using a Nikon Confocal Microscope with the NIS-Elements software, 37˚C, 5% CO2 at 10 X magnification using DiI lasers (em: 595 nm). To provide quantitative results the confocal images were loaded into the NIS-Elements software (Nikon, UK) where they were analysed using the region of interest statistics function that calculated mean intensity values. To standardise the results obtained from DiSBAC23 staining and reduce the effect of cell density on fluorescence, the cell number in each image was counted, and mean intensity values were divided by the respective cell count.

Dielectrophoresis

THP-1 monocytes and M0 macrophages (differentiated per the protocol above) were centrifuged at 300 g for five minutes, resuspended in dielectrophoresis (DEP) medium (8.5% w/v sucrose and 0.3% w/v dextrose, (Sigma-Aldrich, UK), 250 µM MgCl₂ and 100 µM CaCl₂ (both from ThermoFisher, UK) supplemented with isotonic PBS to obtain a conductivity of either 10, 43, 80 or 103 mS/m measured using a conductivity probe. Cells were centrifuged again, and resuspended in medium of the same composition at a cell density of 1 × 10⁶ cells/mL. Ensemble cell electrical properties were measured using a DEPtech 3DEP (Uckfield, UK) for 30 s at five points per decade (10 kHz—20 MHz). Membrane conductance, membrane capacitance, and cytoplasm conductivity parameters were extracted by fitting DEP spectra to the standard Clausius-Mossotti model using MATLAB (Mathworks, Natick, USA). Cell size was determined using ImageJ to measure the radius along the major axis of 100 cells per sample.

Ion gradient potential measurements

To measure any electrical potentials arising from the ion gradients, open-circuit potential (OCP) measurements were performed using a two-compartment electrochemical setup consisting of separate 15 mL tubes containing the electrolyte solutions of interest, connected to the input of a loaded gradient chip. Ionic continuity between the tubes was established using a salt bridge formed by a strip of filter paper saturated with the same solution as the experimental media. Measurements were performed using a two-electrode configuration consisting of a platinum working electrode, immersed in the higher-concentration compartment, and a reference electrode immersed in the lower-concentration compartment. Potentials were recorded using a potentiostatic time-based protocol (chronopotentiometric mode) implemented in NOVA software, with the system operated under open-circuit condition (0 A) and data acquisition supported by the ECAT Complete Package (Metrohm Autolab, Netherlands). Each system was allowed to equilibrate for 300 s prior to measurement, during which the OCP was observed to ensure stabilization of the electrochemical interface, and the stabilized OCP values were extracted. All reported potential were corrected against the control to isolate the contribution of the concentration gradient.

Results and discussion

Ion gradients bias cell migration, independently of biasing migration

To evaluate the effect of ion gradients on macrophage migration, we used microfluidic gradient chips (Fig. 1a, b) and imposed selective gradients. The trajectories of cells in the different gradients (1 µM/µm K⁺, 2 µM/µm Na⁺, 2 µM/µm Cl⁻, plus control), can be seen in Fig. 2. In these images, the trajectories of cells recorded across the analysis regions are repositioned so as to start at the centre of the circle, allowing the net direction of travel to be more readily observed. The horizontal line denotes the difference between cells moving in the direction of increasing concentration (“up” on the image) and those moving in the direction of decreasing concentration (“down” in the image). In particular, we noted that there appeared to be a strong bias visible in the K⁺ and Na⁺ in the direction of increased concentration; that is, more cells ended above the line at the end of their trajectories than below the line.

Fig. 2.

Macrophage migration appears to show directional bias in ion gradients. Plots show the trajectories of cells tracked in the analysis region in Fig. 1, translated to start at the same point at the centre of each circle. Each figure presents data acquired from a single experiment. The horizontal line between those above it (indicating movement in the direction of increasing concentration) from those below (movement in the direction of decreasing concentration) in three ion species (K⁺, Na⁺, Cl⁻) plus control. Traces not to scale

To quantify directional bias, we analysed net cell migration (comparing the end position of each analysed cell at the end of the experiment to its position at the start) and grouped them in four 90° quadrants, as shown in Fig. 1c. These corresponded to the migration towards the inlet (“in”) and outlet (“out”) ports, and “up” (towards increasing concentration) and “down” (towards decreasing concentration) the ion gradient. The mean and SD of the percentages of cells moving in the direction of each quadrant between the start and end of the analysis period is shown in Fig. 3. It is assumed that in the absence of any gradient or flow, migration would be expected to be randomly biased, with 25% of cells in each quadrant. However, in all migration studies including control, we observed a small drift along the direction of fluid flow (that is, more cells were moving “out” than “in”), suggesting that medium flow exerted a small but observable effect on cell migration. For example, the Control cells exhibited a small extra percentage drift in the direction of flow, with 31.3 ± 12.6% finishing in the “out” quadrant, compared to 24.1 ± 8.8% in the “in” quadrant, 25.3 ± 6.6% “down” and 19.4 ± 2.7% “up” respectively.

Fig. 3.

Ion gradients bias macrophage migration along the gradient axis. Radar plots showing the percentages of macrophages migrating along the flow (x-axis) and gradient (y-axis) directions, showing mean and standard deviation of the values in the four quadrants indicated in Fig. 1c. Ion concentration is in the direction of positive y, flow is in the direction of positive x. Gradients were imposed in three ion species (K⁺, Na⁺, Cl⁻)

When we looked at migration in ion gradients, we found that the direction of migration was considerably altered compared to control, with elevated travel in the y-axis (“up” and “down”) and diminished travel along the x-axis (“in” and “out”), as seen in Fig. 3. To understand this better, we looked for significant differences between motion in the y-axis vs. motion in the x-axis. When we compared these for each case, we only identified statistical significance in the Na⁺ gradient (Fig. 4a). To compare the degree to which cell mobility was biased along the gradient axis, we measured the obliquity of the cell motion by dividing the proportion of all cells in the “up” and “down” quadrants by the proportion finishing in the “in” and “out” quadrants (Fig. 4b). If motion were entirely random, we would anticipate a value around 1 (where movement in the y-axis is as likely as movement in the x-axis). When we investigated the control cells, we found an obliquity of 0.82 ± 0.16 due to the slight bias in the direction of flow, as expected. When we analysed the three gradients, we found values of 1.43 ± 0.43 for K⁺, 1.53 ± 0.65 for Cl⁻ and 2.01 ± 0.65 for Na⁺. When we compared the proportion of cells in the “in” and “out” quadrants against the total of “up” and “down” quadrants using a two-tailed t-test, we found that only Na⁺ showed statistical significance (0.0053), whilst both K⁺ (0.069) and Cl⁻ (0.099) were near significance. Finally, we compared the x-direction and y-direction migration of each gradient normalised against control to account for flow effects (Fig. 4c), we found that all three gradients exhibited significance. These results strongly suggest that ion gradients polarise cell motion to move along the axis, though do not necessarily define the direction of travel (positive or negative) along that axis. In this regard, the observed behaviour is similar to that observed when macrophages encounter, and then migrate preferentially along nanostructured grooves [28]. To explore this further, we needed to consider direction as well as axis.

Fig. 4.

Ion gradients significantly bias migration. a The percentage of macrophages migrating in the quadrants in the x (hatched bars) and y (solid bars) directions in Fig. 2. b The obliquity was determined by dividing the percentage of cells moving along the gradient by the percentage moving along the chip. c The percentage of cells migrating across vs. along the chip, for the three gradient conditions normalised against the mean values for Control; in this case, all three ion gradients exhibit significance*<0.05, **<0.01

Cation gradients dominate the direction of macrophage migration

To analyse how cells interacted with the ion gradients, cells were categorised into those whose final position was closer to the high-concentration side than their starting position (“higher”) or closer to the low-concentration side than their starting position (“lower”), as illustrated in Fig. 1d. We analysed cells in both the larger and smaller, central boxes in Fig. 1b. Results are shown in Fig. 5.

Fig. 5.

Cations bias migration up the ion gradient. Mean and SD of the proportions of macrophages migrating up or down the ion gradient, analysed across the whole chip, with arrows indicating the migration in the direction of increasing concentration (up arrow) or decreasing concentration (down arrow) for the three gradients plus control (Ctrl). Asterisks indicate significance; *<0.05, **<0.01, ****<0.0001 (a) Cells analysed across the larger box in Fig. 1b; (b) cells in the central smaller box in Fig. 1b

Considering K⁺ first, the percentage of cells migrating to higher concentration was significantly (p = 0.038) higher than the number migrating towards lower concentration, across both the larger box (p = 0.038) and central zone (p > 0.0001) in Fig. 3b. Cells appeared to be unwilling to move down a K⁺ gradient but were as likely to remain at the same K+ concentration (by migrating to the “in” and “out” quadrants) as they were to move up the gradient. This is the opposite behaviour to that reported by Susdorf and Campenot [14], in which neurons can grow processes down a K⁺ gradient, but not up a K⁺ gradient.

There was no significance observed in either Na⁺ or Cl⁻ across the larger box, but Na⁺ showed significant (p = 0.0047) migration in the direction of increasing concentration in the central zone in Fig. 3b. This suggests that whilst all three gradients bias the axis of migration, only the cation gradients influence the direction of macrophage travel.

It is notable that the magnitude of the gradient to which cells respond is small; a cell 20 μm along its main axis will experience only a 10–20 µM change in ion concentration between the two ends of the cell. Furthermore, the magnitude of the difference in concentration at either end of the cell as a proportion of the overall ion concentration is striking; for example, the difference in Na⁺ concentration on either side of a cell as a proportion of average Na⁺ concentration is of the order of ± 0.025%.

It is worth noting that these gradients also featured choline as a cation substitute, and that the observed response may be affected by the choline gradient rather than the K⁺, Na⁺ or Cl⁻ gradient. However, whilst the migration bias in the K⁺ and Na⁺ solutions acted down the choline gradient, no directional bias was observed in the similar-magnitude Cl⁻ gradient. This suggests that whilst choline may be a factor in biasing movement along the gradient axis, it is not responsible for dictating the directional bias along that axis, as observed in K⁺ and (to a lesser extent) Na⁺ gradients.

Membrane potential changes due to ion gradients do not drive migratory behaviour

The mechanism by which cells sense and respond to the extracellular ion gradient is unclear. One potential mechanism may be related to the changing extracellular ion concentration altering the local cell membrane potential V_m. This is described by the Goldman-Hodgkin-Katz (GHK) equation, usually expressed in the form [29, 30]:

1

where P_x is the ion permeability for ion x (either Na⁺, Cl⁻ or K⁺), cyt and med refer to concentrations in the cytoplasm and extracellular medium respectively, and R, T and F have their usual thermodynamic meanings. Since the presence of an ion gradient will change the local concentrations along the gradient axis, it is possible that this creates different values of V_m at either end of the cell along that axis, which could then act as a driver for cell migration by activating migration-modulating ion channels. To test this, we measured the relative V_m of cells in each of the solutions used to generate the gradients (six total) using the V_m-sensitive fluorescent dye DiSBAC. Significance was tested using an unpaired two-tailed t-test. The results are shown in Fig. 6. For K⁺, fluorescence decreased significantly between 5 mM and 15 mM (p < 0.0001) suggesting depolarisation with increased extracellular concentration, in line with expectations from Eq. (1). Lowering extracellular Cl⁻ significantly raised (p = 0.0113) the relative fluorescence of the cells, indicating hyperpolarisation, again in line with Eq. (1). A decrease in extracellular Na⁺ resulted in a small, non-significant (p = 0.3886) decrease in relative fluorescence. Comparing this to migratory behaviour, we find that cells in K⁺ gradients exhibit gradients in V_m and move in the direction of hypopolarised V_m; that cells in the Cl⁻ gradient exhibited no bias in direction despite the presence of a significant difference in V_m between the two concentrations; and that cells in the Na⁺ gradient exhibited no change in V_m, despite exhibiting the greatest value of obliquity (2.0), and biasing migration in the central zone. This range of different behaviours suggests that neither the obliquity of migration nor the biasing of direction of motion is likely to be driven directly by changes in V_m.

Fig. 6.

Directional bias in macrophage migration is not dependent on membrane potential. Mean (n = 3) fluorescence from cell suspensions treated with the Vm-sensitive fluorescent dyes. Cells were tested at the extreme values of the gradients in (a) K⁺, (b) Na⁺ and (c) Cl⁻ and measured using confocal microscopy. Asterisks indicate significance; *<0.05, ***<0.001

Cell migration in ion gradients may explain electrotaxis

Electrotaxis – the movement of cells in response to an external electrical field – has been reported in many cells [24, 31–38]. It was first reported for mammalian cells over a century ago [34], when migratory leukocytes were observed to move towards the cathode when a DC current was applied. Since then, it has been reported in cancer cells [35], fibroblasts [36] and epithelial cells [37], among others. Most cells (including macrophages) migrate towards the cathode, but some (including monocytes, the macrophage precursor cells) migrate towards the anode [25]. The mechanism by which the cell detects and interacts with the electric field has not been elucidated; the “electric field sensing molecule” has been sought for some time but remains undetected [39]. The pursuit of a mechanism by which cells sense an electric field directly neglects the possibility that the cell may interact with phenomena arising from the field, rather than to the field itself.

Cells are electrically polarizable objects with capacitance and conductance. Whilst an unpolarizable object would not interfere with an electric field (Fig. 7a), cells polarize due to the capacitive components of membrane and cytoplasm which manifests as a displacement of electrical charges and formation of an electrical dipole. The orientation of polarization will depend on the relative electrical resistance and capacitance of the cell and suspending medium; specifically, whether the cell membrane is electrically insulating (Fig. 7b) or conducting (Fig. 7c). Although membrane conductance is very low in viable cells, the effective net cell conductance is a function of both transmembrane conductance, and tangential conductance around the cell surface through the electrical double layer. This is primarily dependent on fixed surface charge attracting conducting ions from the extracellular medium but has been shown to be altered by V_m capacitively inducing charge on the cell surface [40]. The induced dipole potential Ψ_dipole at the poles, due to the electric field E for a spherical entity of radius r, can be determined from the expression [41]:

Fig. 7.

Cells polarize in an electric field. If a non-polarisable object (or one with the same polarisability as the suspending medium) is suspended in an electric field, it does not interact with the field (a). However, if the object is polarisable, it will interact with the field and form a dipole; the orientation of the dipole (red signs) will depend on whether the object is (b) more or (c) less conductive than the medium. d If a cell is suspended in a uniform electric field, the induced dipolar charges will interact with the field, elevating charge on one side and depleting it on the other (cations are shown in this example). This will lead to a cation gradient across the cell (e) (the anion gradient would be the reverse of this). Consequently, it is hypothesised that placing the cell in an imposed ion gradient (f) will have the same effect as an imposed electric field

2

Where σ_p and σ_m are the conductivities of particle and medium respectively, and E is the electric field strength. The orientation of the induced dipole depends on whether the cell is more (σ_p > σ_m) or less (σ_p < σ_m) conductive than the surrounding medium [41]. This dipole comprises induced charges at either pole, along the electric field lines [41]. In accordance with the Poisson-Boltzmann model [42], ions with the same electrical charge (coions) will be repelled from charges on the surface, whereas those of opposite charge (counterions) will be attracted, as described using Eq. (3):

3

Where c_i(0) is the surface concentration of ion species i, c_oi is the bulk concentration, z the valence, e the electronic charge, k the Boltzmann constant and T the temperature. Ψ_cell is the surface potential of the cell. If the dipole potentials could add to (or subtract from) the fixed surface charge at the site of the dipoles, such that Ψ_cell ± Ψ_dipole replaces Ψ_cell in Eq. (3) then this suggests that there are different anion and cation concentrations on either side of the cell; it would induce a local ion gradient across the whole cell, as shown in Figs. 7d, e. Since the ion gradient is induced by the cell itself, it will remain with the cell as it moves. This means that from a cellular perspective, the effect of imposing an electric field on extracellular ion concentrations (Fig. 7d) is equivalent to the presence of an extracellular ion gradient (Fig. 7f).

Using Eqs. (2) and (3) to determine ion concentration at the poles, we find an induced potential difference due to the induced dipole of ± 0.5mV generates an ion concentration difference equal to 8.7% of the concentration in the bulk medium; if the dipole is ± 3 mV, this rises to 52% of medium concentration. The latter example is representative of the case where a 150 Vm^− 1 field (not uncommon in electrotaxis studies) is applied to a cell with 10 μm radius, and σ_p >> σ_m. Assuming a 5mM concentration in the bulk (such as the K⁺ levels found in plasma), for such a cell these values would correspond to concentration gradients across the cell equivalent to 21.8 µM/µm and 131.0 µM/µm respectively.

This approach may explain why cells respond to an imposed field by migrating in opposite directions. If electrotaxis arises from ion gradients generated by induced dipoles, then cells will migrate in opposite directions if the sign of the bracketed term in Eq. (2) differs, causing the dipole to orient as Fig. 7b or Fig. 7c. This in turn depends on relative magnitudes of σ_p and σ_m suggesting different values of cell conductance can orient migration in different directions.

A measurable correlate of σ_p is the effective membrane conductance G_eff. Changes in G_eff due to alteration of cell state have been shown to alter dipole direction; for example, infection of macrophages by Mycobacterium Tuberculosis alters G_eff sufficient to reverse relative polarisability from negative (control) to positive (infected) [43]. This is the equivalent of changing the induced dipole from that shown in Fig. 7c and b and would result in a reversal of the dipole direction and, per our model, changing the direction of migration. Macrophages electrotactically migrate towards the cathode, the macrophage precursor cell, monocytes, migrate towards the anode [24]. We attempted to investigate monocyte migration but were unable to achieve sufficient cell adherence to the chamber surface (monocytes are suspension cells). However, we used dielectrophoresis (DEP) to measure G_eff of undifferentiated THP-1 monocytes and differentiated macrophages at a range of medium conductivities, as shown in Fig. 8. In all cases, G_eff of monocytes was significantly higher than that of macrophages (p > 0.01); this difference was observed to grow with increasing medium conductivity, and at 103 mS/m the value for monocytes was approximately 4x higher than that for macrophages. G_eff has been shown to be linearly related to medium ion concentration up to physiological concentration [44]; fitting to the data here for monocytes suggests a linear correlation with R² = 0.999, whilst for macrophages R² = 0.969, confirming this trend. Extrapolating these linear relationships to physiological ionic strength, the difference between the two would be approximately x6; for comparison, the difference that gave rise to different directions in polarisation direction between infected and uninfected macrophages was just over x2. This suggests that in physiological media, there is sufficient difference in G_eff to generate dipoles in opposing directions in these two cell types, explaining the difference in migratory behaviour between them.

Fig. 8.

Differences in polarizability may explain different electrotactic behaviour of macrophages and monocytes. The mean (n = 3) effective conductance G_eff of monocytes and macrophages in media of conductivity (a) 43 mS/m, (b) 80 mS/m and (c) 103 mS/m, as measured using dielectrophoresis. **, p > 0.01

Other researchers have described factors which influence direction of electrotactic migration, but these may also affect cell polarisability. For example, Kobylkevich et al. [45] used low-density polymers to affect the viscosity, which caused a reversal in migration direction; however, it has been shown by Bondar et al. [46] that polymers in the medium alter extracellular surface potential (and hence G_eff) of cells. Since the ion concentration at the cell surface (with or without a dipole potential) is directly related to the surface potential, it is reasonable that this would alter the cell polarisability (this is before considering that the addition of polymers will alter the medium permittivity, lowering its relative polarisability). Similarly, extracellular pH has been linked to alteration of cell migration direction [47], but pH has been shown to affect both V_m and intracellular K⁺, leading to depolarisation and decreasing relative polarisability [48]. Finally, neurons have been observed to migrate in different directions when cultured on different substrates with different charge polarities [49]. These surfaces are also associated with alterations to the neural phenotype, including its electrophysiology [50, 51] whilst also suggesting that sensing surface charge may represent a variation in the electrical sensing discussed here, by altering ion concentrations at the cell surface.

It is also important to note that ion gradients can generate electrical potentials due to differential diffusion rates of different ion species. To assess whether this effect was present in our system and may have inadvertently caused electrotaxis through a gradient-induced electric field, we measured the electrical potential between the solutions (measured at the inlets), normalised against control. We found that the largest potentials were caused by Cl⁻ gradients (18.6 ± 11.8 mV) and Na⁺ gradients (18.4 ± 10.2 mV) with K⁺ lower (13.6 ± 16.3 mV) reflecting the smaller concentration gradient for that ion. If these are considered across the width of the chamber, this yields a maximum electric field of 2.3 Vm^− 1 for Cl⁻ and Na⁺ solutions, and 1.7 Vm^− 1 for K⁺. As a consequence, we can consider that any movement in our experiments is more likely due to the interaction with the ion gradient, rather than direct sensing of the field itself.

Discussion

There are good reasons to believe that migration in response to detection of K⁺ gradients may be important for macrophage function. Elevation of local K⁺ concentration is associated with cell lysis and cytosolic leakage [52] in conditions such as tumour lysis syndrome [53], brain injury [54] or rhabdomyolysis [55]. Elevated K⁺ is also associated with the tumour microenvironment, where it is known to affect the function of macrophages [16] and T-lymphocytes [56]. In the inflammatory phase after injury, macrophages proliferate within the wound in the days after the injury occurred [57, 58]; elevated K⁺ in normally low-K⁺ bodily fluids is associated with injury [15], and further potassium is secreted by platelets in the clotting process [59, 60]. This elevation in local K⁺ and its associated ion currents may explain the “wound potential” phenomenon, where the potential is a byproduct of elevated ion concentrations and concomitant ion gradients in the vicinity of the wound, which is known to drive the immune response to injury [61].

We considered possible mechanisms by which the cell senses the ion gradient. Several ion channels are associated with macrophage migration; we applied a process of elimination to identify potential candidates for future exploration. The Ca²⁺ concentration is uniform across the chip, suggesting Ca²⁺-activated channels are not responsible [18, 19]. We do not see any mechanisms for the induction of a pH gradient in the chip, suggesting acid-sensing Na⁺ channels did not drive the process [23]. Our analysis of the correlation between changes in membrane potential and migratory behaviour suggests that voltage-sensitive K⁺ channels [18] are also not involved. Similarly, nonspecific cation TRP channels such as TRPV2 [62, 63] are transient and activated by ligands that are unlikely to be present in the experiments described here.

Connexin-43 has also been identified as regulating cell migration [64] in neurons [65] neural crest cells [66], astrocytes [67] and breast epithelial cells [68], whilst elevated levels are also associated with enhanced tumorgenicity and metastatic potential in cancer [69, 70] as well as both migration and wound healing in fibroblasts [71]. This suggests that connexin-43 may play a role in the detection of ion gradients and responding to extracellular ionic conditions. The mechanism by which connexin-43 mediates migration is unknown, and may not be related to its function as a channel, instead acting as a mediator of extracellular signals to the intracellular machinery [72].

Of the other channels have been identified as conferring directionality in migration, inward-rectifying K⁺ (Kir) channels such as Kir4.2, Kir7.1 [73] and other inward-rectifying K⁺ channels [74–77] have been shown to play an important role in electrotaxis. Knockdown of these channels depolarised V_m as well as removing the directional bias of migration without reducing the average migratory distance travelled. Whilst Kir4.2 and Kir7.1 are absent from macrophages, Kir2.1 is present [77]. The role of a specifically rectified channel is significant in the case of biasing channel behaviour up a K⁺ gradient. If the ion concentration were to form a gradient across the cell (as postulated for induced dipoles), altered extracellular concentration would only lead to an inward current on the side where K⁺ is elevated; on the side where K⁺ is reduced, rectification would prevent ion flow. If the leak current through the channel elevates cytoplasm K⁺, then the cell would stop moving once the extracellular concentration equilibrates with the intracellular value.

No rectified Na⁺ channels have been identified in macrophages, whilst the chloride channel ClC3 identified in macrophages [78] has been shown to allow ion transport in both directions, but with different permeabilities. This rectification element is missing in connexin-43, which may explain why Cl⁻ and (to a lesser extent) Na⁺ bias migration along the gradient axis in either direction, such that the cells respond to both elevated and depleted differences between intracellular and extracellular ion concentrations.

This work suggests that ion gradients may represent a form of extracellular signalling beyond more conventional macromolecular messaging by hormones, neurotransmitters or exosomes. Ion gradients have proven a key component in intracellular communication between bacteria both inside and outside biofilms [11–13]. We have shown for the first time that macrophages respond to ion gradients by biasing direction axis, with cation gradients, predominately in K⁺, also defining overall direction. Furthermore, we use the idea of induced ion gradients to suggest a potential mechanism for electrotaxis. Examination of channel expression in cells exhibiting electrotaxis indicate that such cells are likely to express both connexin-43 and rectified K⁺ channels. This suggests that ion gradients may play a functional role in mammalian cell communication and extracellular sensing, particularly in cell migration as well as other electrically-mediated cellular phenomena including neural growth, cell division, and cell behaviour in cancer.

Authors’ contributions

OVG and MKC performed experiments and analysed the results. MPH developed the theory, supervised the project and analysed the results. HAK performed analysis. SC, RL and FHL provided scientific knowledge and guidance. All authors contributed to writing the paper.

Funding

This work was partially funded by Khalifa University grant KU-2022-020 to MPH, and UAEU grants G00005469 and G00004633 to FHL. The work has been supported by the Khalifa University Biotechnology Center.

Data availability

Data are available on request from Michael Pycraft Hughes, michael.hughes@ku.ac.ae.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

Author MPH is co-inventor of the 3DEP instrument used in part of this work and a director of the company that manufactures it. The other authors declare no conflicts of interest.