Skip to main content

The retinal pigment epithelium displays electrical excitability and lateral signal spreading



The non-neuronal retinal pigment epithelium (RPE) functions in intimate association with retinal photoreceptors, performing a multitude of tasks critical for maintaining retinal homeostasis and collaborating with retinal glial cells to provide metabolic support and ionic buffering. Accordingly, the RPE has recently been shown to display dynamic properties mediated by an array of ion channels usually more characteristic of astrocytes and excitable cells. The recent discovery of canonical voltage-activated Na+ channels in the RPE and their importance for phagocytosis of photoreceptor outer segments raises a question about their electrogenic function. Here, we performed a detailed electrophysiological analysis related to the functioning of these channels in human embryonic stem cell (hESC)-derived RPE.


Our studies examining the electrical properties of the hESC-RPE revealed that its membrane mainly displays passive properties in a broad voltage range, with the exception of depolarization-induced spikes caused by voltage-activated Na+ current (INa). Spike amplitude depended on the availability of INa and spike kinetics on the membrane time constant, and the spikes could be largely suppressed by TTX. Membrane resistance fluctuated rapidly and strongly, repeatedly changing over the course of recordings and causing closely correlated fluctuations in resting membrane potential. In a minority of cells, we found delayed secondary INa-like inward currents characterized by comparatively small amplitudes and slow kinetics, which produced secondary depolarizing spikes. Up to three consecutive delayed inward current waves were detected. These currents could be rapidly and reversibly augmented by applying L-type Ca2+ channel blocker nifedipine to diminish influx of calcium and thus increase gap junctional conductance.


This work shows, for the first time, that INa and INa-mediated voltage spikes can spread laterally through gap junctions in the monolayer of cells that are traditionally considered non-excitable. Our findings support a potential role of the RPE that goes beyond giving homeostatic support to the retina.


Retinal pigment epithelium (RPE) is adjacent to photoreceptors in the eye and critical for normal retinal functioning. RPE secretes growth factors; phagocytoses the exhausted fragments of photoreceptor outer segments (POS); regenerates the visual pigment; maintains ionic homeostasis of the sub-retinal space by transporting ions, water, and metabolic end products between the sub-retinal space and the blood; serves as a barrier between the bloodstream and the retina; absorbs light; and protects the retina from photo-oxidation [1, 2]. A loss or failure of any RPE function can have pathologic consequences in the retinal tissue leading to photoreceptor degeneration and even blindness [2, 3]. Several of these functions depend on ion channels expressed in the plasma membrane [2, 4,5,6,7,8,9,10] .

The channelome of RPE is large and includes many ligand- and voltage-activated channels with functional verification to at least TRP (TRPC1, 4; TRPV2); inwardly (Kir4.1, 6.2, 7.1) and outwardly rectifying voltage-activated K+ channels (KV1.2, 1.3, 1.4, 4.2; KCNQ1, 4, 5); Ca2+-activated K+ channels (BK); voltage-activated Ca2+ channels (CaV1.2, CaV1.3, CaV3.1, CaV3.2); Cl channels (ClC-2, best-1, cystic fibrosis transmembrane conductance regulator (CFTR)); connexin channels; and voltage-gated Na+ channels [5,6,7,8, 11,12,13,14,15,16,17,18]. Yet, RPE cells are not neurons; they are traditionally considered to have passive intrinsic properties, similarly to other non-excitable cells. Is this large channelome with several voltage-activated channels present in the RPE only to support its homeostatic activity?

This study was motivated by the question raised above as well as the recent discovery of voltage-activated Na+ channels in the RPE and their prominent role in the phagocytosis of POS [4]. Immunolabeling and mass spectrometry experiments revealed the expression of nine alpha-subunits of voltage-activated Na+ channels (NaV1.1, NaV1.3 to NaV1.9, and NaX), especially NaV1.4, NaV1.6, and NaV1.8 [4]. It was shown that exposure of RPE to tetrodotoxin (TTX) both in culture and ex vivo in the mouse, and also silencing the expression of NaV1.4 with shRNA, can suppress phagocytosis, raising a possibility that electrogenic function of these channels might contribute to phagocytosis [4]. By demonstrating the presence of Na+ channels also in the mouse RPE, these results contradict conclusions of earlier studies, in which inward Na+ current (INa) was observed but attributed to the neuronal trans-differentiation of cultured primary RPE [16, 19].

Fast depolarizing voltage responses (“spikes”) mediated by INa have been previously described in the conventional non-excitable cells, including astrocytes and rabbit pigmented ciliary body epithelial cells [20,21,22,23]. In the study of cultured neonatal rat RPE, Botchkin and Matthews reported action potential-like depolarizing spikes [16]. However, although many studies have analyzed ionic currents in various RPE cells, our understanding of membrane properties and voltage responses is still incomplete.

Development of functionally mature human embryonic stem cell-derived RPE (hESC-RPE) holds promise for prospective transplantation treatments of RPE-dependent retinal degenerative diseases [24,25,26]. The hESC-RPE cell lines express RPE-specific genes and proteins, exhibit a hexagonal cobblestone morphology with appropriate polarization and barrier properties, strong pigmentation, and perform phagocytosis similarly to human fetal RPE [5, 27,28,29,30]. This functionality and promising outlook of hESC-RPE transplantation necessitate comprehensive study of their electrophysiological properties. In this study, by using the patch-clamp technique, we investigated membrane response properties of hESC-RPE cells in the monolayer in physiological conditions at body temperature. We describe the properties of spikes, a spontaneously and rapidly fluctuating leak conductance, and a previously unreported phenomenon of multiple waves of inward currents caused by signal spreading via gap junctions.


Ionic currents in hESC-RPE

We studied electrophysiological properties of pigmented, cobblestone hESC-RPE cells (Fig. 1A) in whole-cell patch-clamp experiments by recording from 707 cells in voltage clamp mode; in 181 of these cells, additional current-clamp measurements were performed. There was a substantial variability in the ionic currents between cells (Fig. 1C,D). INa of variable amplitude (see INa in Fig. 1C and D for comparison) was detected in the majority (~98%) of cells using the recording protocol shown in Fig. 1B. The transient (IA) and sustained (delayed rectifier, IDR) voltage-activated K+ currents were expressed at dissimilar ratios (Fig. 1C, D). Similarly, leak conductance varied from cell to cell. Sustained hyperpolarization-activated inward current (IIR, Fig. 1E), possibly mediated by K+ inward rectifier channels, was found in the majority (~80%) of cells (Fig. 1D) [6].

Fig. 1
figure 1

Ionic currents in hESC-RPE and their variability. A Bright-field microscopy image of hESC-RPE monolayer with the patch pipette attached to the cell. Scale bar is 10 μm. B Experimental protocol to record INa; a 50-ms pulse to −80 mV was given from a holding potential of −70 mV to measure leak current and membrane capacitance; a 50-ms pre-pulse to −170 mV was used to recover INa before stimulating with 50-ms testing pulses in 10 mV increments from −80 to 100 mV. C, D Examples of whole-cell currents recorded from two cells; IC, residual capacitive current; INa, voltage-activated Na+ current; IA and IDR, the transient and sustained outward voltage-activated K+ currents, respectively. E An example of sustained hyperpolarization-activated inward current (IIR)

The ultra-fast transient depolarizing INa is conducted by several Na+ channel types [4]. Figure 2 describes the essential aspects of INa in hESC-RPE at the physiological temperature of 33–35°C. A typical INa is shown in Fig. 2A. Its amplitude was determined by subtracting the current after the end of the preceding capacitive current (IC) from the negative peak amplitude. At positive voltages, precise measurement of INa was often complicated due to progressive superposition of the partly compensated and slightly fluctuating IC with INa as the activation kinetics of INa accelerated with depolarization. However, this complication was usually not present at voltages below 0 mV. INa activation threshold was in between −40 and −30 mV. The average current-voltage (IV) relation for a group of cells without capacitive artifacts is shown in Fig. 2B. It can be seen that the current disappears by +80 mV, consistently with the +86 mV reversal potential for Na+ created by the recording solutions (see “Methods”).

Fig. 2
figure 2

Kinetics of voltage-activated Na+ currents (INa). A Typical INa; IC, residual capacitive current; IK, the putative voltage-activated K+ current. B Average current-voltage (IV) relationship for INa (n = 25); here and elsewhere error bars are s.d. (Additional file 2: Table S1). C Steady-state inactivation (SSI) was studied by evoking INa at 0 mV after 50 ms conditioning pre-pulses from −150 to −10 mV. D Voltage dependencies of the average peak INa (circles) and its time to peak (triangles) from SSI experiments (n = 17), and voltage dependence of the normalized peak INa conductance (n = 25, squares). SSI data were divided by the amplitude of INa after the pre-pulse to −150 mV; in the figure, average values were fitted with a sigmoidal equation. Conductance at each membrane potential V was calculated using peak INa from IV relations as in B and equation gNa(V) = INa/(VErev), where ENa = 85.8 mV (Additional file 2: Table S1). E Recovery of INa from inactivation. INa was evoked and inactivated by 50-ms pre-pulses to 0 mV, after which hyperpolarizing pulses of increasing duration ranging from 0.5 to 21.0 ms were applied to recover INa; this was followed by a testing pulse to 0 mV. In these experiments, the holding potential between the trials and the recovery potential were the same. F By plotting peak INa during recovery from inactivation against the hyperpolarizing pulse duration and then fitting the data with a first-order exponential rise-to-maximum equation, recovery time constants (τrec) can be obtained; examples are from the same cell as in E. G Voltage dependence of average recovery time constants (Additional file 2: Table S1)

In addition to INa, RPE can express voltage-activated Ca2+ channels that mediate an inward Ca2+ current with voltage dependencies similar to those of INa but slower kinetics [31, 32]. Because calcium currents in hESC-RPE appear to be very small and mostly non-inactivating, requiring replacement of Ca2+ ions by Ba2+ ions in the external solution for measurement [5], we did not investigate them specifically. However, we verified the consistency of INa characteristics without the contribution of K+ and Ca2+ channels using Cs+-based intracellular solution and 10 μM nifedipine extracellularly (Additional file 1. Fig. S1) [33].

Next, we determined voltage dependencies of several INa parameters. Availability of INa was investigated using a steady-state inactivation (SSI) protocol consisting of 50 ms conditioning pulses between −150 and −10 mV followed by a test pulse to 0 mV (Fig. 2C). As can be seen from the gradually decreasing INa during the test pulse, exposure to voltages in the range from ca. −120 mV to the INa activation threshold caused closed-state inactivation (Fig. 2C, D). The half-inactivation potential was −57.8 ± 4.4 mV (Fig. 2D). In particular, 27% of INa was on average available after a brief exposure to −50 mV and 11% after holding at −40 mV. Interestingly, INa time to peak decreased steadily with increasing voltage of the conditioning pulse, from 1.60 ± 0.30 ms after holding at −150 mV to 1.27 ± 0.21 ms after holding at −70 mV (P = 10−11, paired t-test, Fig. 2C, D), indicating voltage-dependent facilitation of INa. This is probably caused by accumulation of Na+ channels in the conformation states that promote both rapid opening and steady-state inactivation. The overlap between the voltage dependencies of activation (obtained from IV relations, the half-activation potential of 23.1 ± 1.6 mV) and inactivation indicated that a small window current might exist around −35 mV (Fig. 2D).

Recovery from inactivation was evaluated at four membrane potentials (Fig. 2E–G). After a 50-ms pulse to 0 mV to activate and inactivate INa, the cell was exposed for variable durations to the test potential, from 0.5 to 21.0 ms in 0.5-ms increments. The resulting response series allowed measuring the time constant of recovery by fitting the time course of peak INa increase with a first-order exponential rise-to-maximum equation (Fig. 2E, F). Time courses of recovery with exponential fits are shown for one cell in Fig. 2F. Figure 2G demonstrates the average voltage dependence of recovery time constants. INa recovered with the time constant of 1.05 ± 0.26 ms at −120 mV but only 16.9 ± 6.7 ms at −60 mV.

Spikes in the current-clamp experiments

The presence of INa in hESC-RPE raises an immediate question about the capability of RPE cells to generate transient depolarizations, such as action potentials or spikes. To further investigate their electrophysiological properties in general and voltage spiking in particular, we used several current-clamp protocols, all of which contained a hyperpolarizing current injection to recover INa prior to a test stimulus. The duration of such hyperpolarizing pulses was set to 50 ms and amplitude varied depending on the membrane resistance in order to hyperpolarize membrane potential to below −100 mV to quickly recover most of INa (Fig. 2G).

Figure 3A shows ionic currents and Fig. 3B voltage responses to a series of depolarizing current pulses recorded in the same cell. Spikes appeared when membrane was depolarized to above −40 mV, with their times to peak decreasing with depolarization. Another important feature was that in the absence of spikes, the RPE membrane behaved like a passive electrical circuit, i.e., the time course of charging could be easily fitted with a first-order exponential function (see the stable membrane resistance in Fig. 3E).

Fig. 3
figure 3

Fast sodium spikes in current-clamp experiments. A Example of voltage clamp recordings. B Spikes in a current-clamp experiment were evoked by first hyperpolarizing the cell with 100-ms current injections to below −100 mV and then applying depolarizing pulses as indicated. C Separation of a spike (green trace) from the voltage response (black); the region containing a spike was excised and the remaining trace fitted with a first-order exponential rise-to-maximum equation (red); subtracting the traces elicited the spike. Membrane resistance (Rm) was determined from the amplitude of the voltage response (ΔV) to the injected current (Iinj); membrane time constant (τm) obtained by the fitting was used to calculate membrane capacitance (Cm) as indicated. D Examples of spikes; asterisks correspond to the marked traces in B. E Voltage dependencies of membrane resistance, spike amplitude, and spike time to peak in the same experiment; note that the membrane potential refers to the steady-state voltage reached after membrane charging is finished, which is in most cases higher than the potential at which the spike is initiated. F Voltage dependencies of spike parameters: half-widths, 25–75% rise and 75–25% decay times

To study the properties of spikes, it was necessary to separate them from the underlying voltage responses. We used a procedure described in Fig. 3C. The spike distorts the rising phase of the voltage response causing a deviation from the uniform exponential rise to maximum. By removing the spike-containing part of the trace and fitting the rest with a first-order exponential rise-to-maximum equation, an estimate of a response without the spike can be obtained (Fig. 3C, red trace). By subtracting it from the original trace, the spike waveform is acquired (Fig. 3C, green trace). Spikes from Fig. 3B are shown in Fig. 3D and their properties are described in Fig. 3E, F. Spike amplitude changed little with depolarization, unlike the amplitude of INa (Fig. 3E). In contrast, spike time to peak decreased strongly with depolarization, similarly to changes in the time to peak of INa (Fig. 3A), albeit in a much wider temporal range due to the relatively slow membrane charging in the current-clamp experiments. Kinetics of spikes, including the 25–75% rise and decay times, also accelerated, resulting in a gradual narrowing of spikes with depolarization (Fig. 3F). Because membrane was passive in the studied voltage range, these differences could be attributed to voltage dependence of INa kinetics.

Although evidence suggested that spikes are caused by INa, it was necessary to perform a definitive test. We used 10 μM TTX to block INa and evaluate changes in spikes (Fig. 4). TTX strongly suppressed INa (Fig. 4A). Simultaneously, spike amplitudes decreased dramatically and clearly in proportion to changes in the peak INa (Fig. 4B, C). This was observed in all three experiments (P = 0.02 for both INa and spike amplitude, paired t-test, Fig. 4D). These results unequivocally prove that spikes are mediated by INa.

Fig. 4
figure 4

Effect of TTX on INa and spikes. A IV relations from an experiment involving inhibition of INa with 10 μM TTX. B Traces of current-clamp recordings in control conditions with spikes visible and after application of TTX from the same cell as in A; data were obtained as described in Fig. 3B; stimuli not shown. C Isolated spikes from B. D Summary of TTX experiments. Amplitudes of the largest INa and spikes were measured in three cells before and after application of TTX; values in the presence of TTX were divided by control ones and averaged (Additional file 2: Table S1)

Next, we investigated the voltage dependence of spikes. Figure 5A–C shows a representative experiment. The cell was stimulated with 100 ms hyperpolarizing conditioning current pulses of different amplitudes to obtain membrane potentials in the range from below −120 mV (to fully recover INa) to about resting potential. The hyperpolarizing pulses were followed by relaxation to resting potential, which triggered the spikes, depending on the preceding membrane potential. The spike amplitude and time to peak decreased with depolarization, whereas the half-width increased (Fig. 5C). The decrease in amplitude was due to closed-state inactivation of INa as the consistent average half-inactivation potential for INa and the half-maximal spike amplitude potential indicate (Fig. 5D). The decrease in the time to peak was due to a comparatively fast membrane charging toward the INa activation threshold from more depolarized conditioning potentials.

Fig. 5
figure 5

Voltage dependence of spikes. A Voltage responses of an RPE cell at resting potential after a series of 100-ms conditioning hyperpolarizing pre-pulses. B Isolated spikes from A, with asterisks indicating the correspondence between spikes in A and B. C Dependence of spike amplitude, time to peak and half-width on the pre-pulse voltage; data from A and B. D Comparison of the average half-inactivation potentials from SSI recordings and the average half-maximal amplitude potentials for spikes (both denoted with V0.5) in the same cells; the voltage dependencies were fitted with a sigmoidal equation (Additional file 2: Table S1).

Because voltage-gated K+ conductances can condition spikes caused by INa, and also because activation of delayed rectifier K+ conductances by a depolarizing current injection can cause spike-like voltage transients [34], we correlated spike sizes and half-widths with the amplitudes of IDR and IA. The currents were measured after a 50-ms pre-pulse to −170 mV at the nominal voltage pulse to +100 mV (see Fig. 1B). IA was measured as the difference between the peak outward current and the sustained current at the end of 50-ms pulses. IDR was determined at the end of 50 ms pulses, with leak subtracted. We found a weak positive correlation (r = 0.2, P = 0.019, n = 140) between IA and spike amplitudes but no correlation between IDR and spike amplitudes. No correlations could be found between IA and IDR amplitudes, on the one hand, and spike half-width, on the other hand.

In contrast, the maximal spike amplitude correlated strongly with the peak INa, with r = −0.66 (Fig. 6A, n = 95). Figure 6B clearly shows that the distribution of peak INa was not Gaussian, with a median of −1140(−765:1635) pA (n = 105).

Fig. 6
figure 6

Spikes, INa and membrane time constant. A Correlation between the maximal spike and INa amplitudes; INa was measured from a recording immediately preceding the current-clamp. B, C Distribution of maximal INa amplitudes (B) and Cm values (C); the distributions are not Gaussian (P < 0.001, Shapiro-Wilk normality test). D Correlation between the maximal spike amplitudes and INa densities; INa density was obtained by dividing peak INa with the corresponding Cm value. E Plots of spike time to peaks and half-widths against membrane time constants; the spikes were obtained from voltage responses reaching the level of ca. 0 mV (−0.9 ± 2.7 mV, n = 60) from voltages below −100 mV (−131 ± 21 mV, n = 60); the correlations were characterized by P values of < 10−6

However, membrane charging by INa during the onset of the spike depends not only on INa amplitude but also on its density. Figure 6C shows that capacitance, a proxy for cell membrane area, varied substantially in our experiments. The distribution was also not Gaussian, with a median of 17.1(11.6:21.9) pF (n = 99). The correlation between the maximal spike amplitude and INa density obtained by dividing peak INa by Cm (Fig. 6D) was −0.72, stronger than that in Fig. 6A.

Speed of membrane charging varied between cells and often in the same cell during prolonged recordings (see next section). To test how variability in membrane time constant could affect the spike, we correlated spike properties to the corresponding time constants. Although membrane time constant (τm) did not depend on membrane potential, INa did, and so in the following analysis we used spikes obtained from voltage response traces with sustained potentials at around 0 mV and evoked from voltages below −100 mV to minimize closed-state inactivation of INa. There was no correlation between spike amplitude and τm (r = 10−2), i.e., the speed of membrane charging had no effect on the spike size. However, a nearly unity (0.98) correlation was discovered between τm and spike’s time to peak (Fig. 6E), and a less strong correlation (r = 0.89) was observed between τm and spike’s half-width (Fig. 6E). The latter correlation was caused by strong dependencies on τm of spike’s 25–75% rise and decay times, with the corresponding correlation coefficients of 0.74 and 0.88 (for both, P < 10−6; n = 60).

Although we did not register spontaneous spikes when membrane potential was not hyperpolarized by current injections, there is evidence that they can be evoked if membrane potential transiently decreases to recover a sufficient fraction of INa. Figure 7A demonstrates three traces recorded in a current-clamp experiment. The red trace contains a spike (arrow) triggered by a sharp spontaneous depolarization after a brief hyperpolarization (two asterisks). The same trace also contains episodes of rapid depolarization (asterisks), which, however, failed to trigger spikes. Immediately after the spontaneous spike, the hyperpolarizing current pulse was switched off and the cell depolarized to about 0 mV (Fig. 7B). Because available Na+ channels were inactivated during the spike, no new spike could be elicited, in contrast to the other two traces (black and green), where spikes did appear.

Fig. 7
figure 7

Stochasticity of spike onset. A Spikes can occur apparently spontaneously when the cell is hyperpolarized by current injection; asterisks denote two particularly sharp depolarizing transients and the double asterisk the region of spontaneous hyperpolarization preceding the actual spike (arrow). B Magnified spike-containing segment from (A). C IV and SSI relations for another cell characterized by a large INa with a comparatively negative INa activation threshold. D Parts of three voltage traces from a current-clamp recording, with a spike at 360 ms on the black trace. E Variability in spike properties in the INa activation threshold region. F Distribution of minimal resting potentials, its Gaussian fit, and the INa availability curve obtained by fitting the SSI data from Fig. 2D; the Vrest histogram data were divided by the maximum value. G, H Ionic currents (G) evoked from a holding potential of −70 mV in an RPE cell with a very negative Vrest and the associated current-clamp recording (H)

In another experiment, INa was large, had a low activation threshold, and its substantial fraction (~20%) was available after a 50 ms exposure to −40 mV (Fig. 7C, white triangles). Figure 7D shows three voltage traces from the associated current-clamp recording. A large spontaneous spike can be seen on the black trace (arrow). Its activation threshold was about −60 mV and, by comparison with two other traces characterized by the similar membrane potential, it appears that the spike was evoked because a substantial fraction of INa recovered during the preceding hyperpolarization between 220 and 310 ms.

Figure 7E illustrates the stochasticity of spike amplitudes and kinetics when they were triggered around the INa threshold potential in another RPE cell. Seven traces are shown. Clear spikes can be seen in four traces during two depolarizing segments. No spike appeared in the black trace, although during the second depolarizing segment a slow transient was evoked. This trace was characterized by the most positive membrane potentials before the depolarizing pulses, so that 97% of the maximal INa was available before the first and 84% before the second depolarizing segment (values from the associated SSI recording). Similarly, no spike was evoked during the red trace characterized by even more hyperpolarized potentials. Green and cyan traces almost coincided with each other and with the red trace in their pre-depolarization membrane potential but only the green trace had clear spikes. Likewise, gray and blue traces were quite similar in their membrane potentials, but spikes appeared only on the blue trace. Two spikes can be seen in the pink trace but their timing relative to other spikes was different during the two depolarizing segments.

Although we found no clear spikes at resting membrane potential (Vrest) in our recordings, any residual INa persisting at Vrest or recovered by a preceding hyperpolarizing fluctuation of membrane potential can plausibly amplify the next rapid depolarizing fluctuation. The resulting transient might resemble the smallest dark gray spikelet in Fig. 5B or the fast depolarizing spikelets on the red and black traces in Fig. 7A. To estimate the available fraction of INa at rest in the experimental group, we plotted the histogram (Fig. 7F) of the most negative Vrest values obtained in each cell in the absence of current injection (e.g., from the green trace in Fig. 7A at 0 ms or the black trace in Fig. 9 at 0 ms). The average Vrest was −29 ± 18 mV (181 cells), which would become −42 mV after the liquid junction potential error correction. The histogram was fitted with a Gaussian equation (blue trace) and superimposed on the SSI fitting curve from Fig. 2D (red trace). In the minority of cells characterized by Vrest below −50 mV, more than 40% of INa could be available. However, in the majority of cells, INa at Vrest was inactivated almost completely.

An example of a cell with a very low Vrest is shown in Fig. 7G, H. The currents were recorded from a holding potential of −70 mV. Figure 7G shows comparatively large INa, IA, IDR, IIR, and a small leak current in the vicinity of Vrest. In the current-clamp measurements, Vrest occasionally reached −80 mV, indicating the availability of a large fraction of INa. Consequently, upon stimulation with depolarizing current injections, spikes were elicited after crossing the INa activation threshold at about −40 mV. The low Vrest was associated with very high average Rm of 1730 MΩ and membrane time constant of 35 ms, which could explain the stability of membrane potential and absence of fluctuations.

Dynamic changes in membrane resistance

Figure 3 shows current-clamp recordings characterized by relatively stable Rm (Fig. 3E) and Vrest values (Fig. 3B). However, such stable recordings were exceptions rather than a rule because Rm and Vrest tended to change continuously. Figure 8A, B demonstrates parts of two recordings from the same cell, both displaying large spikes (Fig. 8C, D). During the recording in Fig. 8A, the cell had relatively small τm (Fig. 8E) and Rm (Fig. 8F), on average 222 MΩ. To charge its membrane potential to below −100 mV, a comparatively large hyperpolarizing current of −500 pA was needed (top).

Fig. 8
figure 8

Spontaneous changes in leak conductance and its effects. A, B A recording with a low Rm (A) was followed by a recording with a relatively high Rm (B); C, D Spikes from the low- (C) and high-resistance (D) recordings; color coding is the same as in A and B, respectively. E Dependencies of membrane time constants and capacitances on the response voltage immediately after the spike. F Dependencies of membrane resistances on the response voltage immediately after the spike; A–F, data from the same cell. G Time courses of changes in Rm (top) and resting potential Vrest (bottom) during one experiment. HK Correlations between Rm and Vrest during similarly prolonged recordings in 4 cells

In contrast, the recording in Fig. 8B was characterized by large τm (Fig. 8E) and Rm (Fig. 8F), and a small −100 pA current hyperpolarized membrane potential to below −100 mV. The relatively low-resistance green trace in Fig. 8B illustrates sudden changes in Rm during the recordings (each trace is a part of a 1-s recording). Importantly, while the average Rm and membrane time constant increased by threefold, the average Cm did not change (Fig. 8E), supporting the validity of our estimation procedure (Fig. 3C). The differences in the properties of spikes (Fig. 8C, D) were consistent with the findings shown in Fig. 6E: changes in membrane time constant had a small effect on the spike amplitude but strongly altered their times to peak, half-widths, and kinetics.

Top plot of Fig. 8G shows an example of periodic changes in Rm over the course of a prolonged experiment consisting of many current-clamp recordings. Rm varied from 60 to 3900 MΩ, which corresponded to changes in membrane conductance from 17 to 0.26 nS. Comparison of time courses of changes in Rm and Vrest (Fig. 8G, bottom) indicated that Vrest depolarized when Rm decreased and vice versa. This implies that the fluctuating membrane leak was depolarizing, i.e., mediated by an inward cationic or outward anionic current. Figure 8H shows Vrest values plotted against the respective Rm values for the whole experiment in Fig. 8G, and Fig. 8I–K demonstrates similar plots for three more experiments. All results were consistent: Vrest was a function of membrane leak.

Although changes in Rm and Vrest usually occurred gradually, they were often facilitated by strong and sudden membrane depolarization or hyperpolarization. Figure 9 shows two examples of rapid hyperpolarization-induced decreases in Rm. Over the course of the black trace, the cell was first characterized by a comparatively negative Vrest and high Rm, >1100 MΩ. Even before the membrane potential had fully settled during a −250 pA current pulse, a sudden decrease in Rm, to ~350 MΩ, happened at 62 ms. Eventually, after the decrease in Rm at ~370 ms, the cell acquired a resistance of 140 MΩ. Accordingly, Vrest increased from −32 mV at the trace onset to +2 mV at 650 ms. The second negative transient, with the corresponding Rm estimates, can be seen on the red trace at 670 ms.

Fig. 9
figure 9

Sudden changes in leak conductance. While leak conductance can change gradually, in some cases its decrease occurs suddenly in the beginning of a hyperpolarizing pulse, manifesting in a sharp hyperpolarizing transient. Rm values in black font refer to the black trace and in red to the red trace

Multiple waves of excitation

In 19 out of 707 studied cells, in addition to the conventional INa, we found anomalous secondary depolarizing currents (Fig. 10A) characterized by relatively large times to peak, with slow onsets and inactivation kinetics. The currents were clearly voltage-dependent.

Fig. 10
figure 10

Multiple inward currents. A Voltage clamp recording with two depolarizing currents, a conventional INa (inward 1) and a relatively small and slow delayed inward current (inward 2). B The IV relations for the two currents and the scaled up inward 2 current; inward 2 current was isolated by (1) removing the delayed current-containing part of the trace, (2) fitting the remaining trace between 5 and 30 ms with a polynomial equation, and (3) subtracting the fitting curve from the original trace. C A current-clamp recording from the same cell with two types of voltage spikes that correspond to the currents in A. DF A recording from another cell, D with three waves of inward currents, E the corresponding IV relations, and F the current-clamp recording. G Example of four inward currents from a different cell. H An example of recordings where the delayed inward current (inward 2, left) disappeared spontaneously after 5 min of recordings (right)

To investigate the properties of these delayed inward currents, we separated them from the outward current using a procedure similar to that we used to isolate spikes (Fig. 3C). Comparison of the IV relations for INa and the delayed inward current showed that the extrapolated reversal potential of the new current should be extremely positive (ca. +200 mV, Fig. 10B). In the current-clamp experiments, the delayed inward current caused secondary spikes (Fig. 10C). Similarly to the closed-state inactivation of INa (Fig. 2D), the delayed inward current is inactivated by pre-exposing the cell to the same range of membrane potentials.

Surprisingly, detailed examination of several recordings characterized by the presence of the second inward current revealed additional delayed inward currents (Fig. 10D–G). In Fig. 10D, the third inward current (inward 3) was small and even more delayed and slower than the second one (inward 2), eliciting small and slow depolarizations (Fig. 10F). In this cell, the second inward current IV relation (Fig. 10E) had apparently a similarly positive reversal potential as the delayed inward current in Fig. 10A, B, whereas the IV relation for the third current could not be reliably estimated beyond +70 mV. It should be noted that the cells with delayed inward currents tended to have unusually high Cm values (see the prolonged membrane charging in Figs. 10A, D, G and 11A–C, G–I), which could not be compensated properly, hence the nominal voltage in IV relations was probably a large overestimation. Membrane capacitance estimated in the voltage clamp recordings from the charge during the capacitive transient was on average 54(29:69) pF (n = 19) in the cells with the secondary delayed inward current but only 26(21:34) pF (n = 685) in the typical cells (P < 10−3, Mann–Whitney U-test).

Figure 10G shows voltage clamp recording from a different cell with four depolarizing inward currents. Three inward currents were present at all potentials, except the threshold one, where the conventional INa and the first delayed inward current partly merged. The inward current 4 in Fig. 10G was small and slow, resembling the Fig. 10D current, and appeared at more positive potentials than the inward currents 1–3.

The presence of multiple depolarizing currents rules out a possibility of their origin in the same cell. This phenomenon can only be explained by the spread of excitation in both voltage- and current-clamp experiments to the neighboring cells via open gap junction channels. Consistently with this hypothesis, we found that the delayed inward current can be regulated dynamically, as shown in Fig. 10H. In six cells, it disappeared completely within minutes after the start of voltage clamp recordings.

Furthermore, gap junction conductance can be increased by lowering internal calcium concentration and vice versa [35]. Because L-type Ca2+ channels are expressed in hESC-RPE and are partly open at their typically depolarized resting potentials [5], we attempted to alter the internal calcium homeostasis by blocking these channels with nifedipine, which at 10 μM abolishes most of L-type Ca2+ channel current [5]. We recorded from five cells exhibiting delayed inward currents. Surprisingly, delayed inward currents increased noticeably immediately after application of 10 μM nifedipine. Figure 11A–C shows the effect of nifedipine in one cell. The effect was fully reversible (Fig. 11C, D). Bar plot in Fig. 11E summarizes the results of the experiments: on average, the maximal delayed inward current increased by more than 2-fold.

Fig. 11
figure 11

Effects of nifedipine and 18α-GA on the delayed inward current. AC Voltage clamp recordings in control conditions (A), after application of 10 μM nifedipine (B), and after wash-out (C). D IV relations for the delayed inward currents from A to C (inward 2). E Bar plot summarizes changes in the peak delayed inward current in 5 cells in the presence of 10 μM nifedipine; paired t-test was used for the statistical comparison. F Cells were first exposed to 10 μM nifedipine and then, after wash-out of nifedipine, to 10 μM 18α-GA; bar plot shows changes in Cm (n = 6; P = 0.002, paired t-test). GJ Voltage clamp recordings in control conditions (G), after application of 10 μM 18α-GA (H), and, after wash-out, 10 μM TTX (I); J IV relations for this experiment. In panels F–J, recordings were performed with Cs+ in the pipette solution

If nifedipine facilitates opening of gap junctions, then this can manifest in changes of Cm. However, because the gap junction conductance status in control cells is not known, we compared Cm in the presence of 10 μM nifedipine to that after wash-out of nifedipine and application of connexin 43 (Cx43) channel inhibitor 18α-glycyrrhetinic acid (18α-GA). Cx43 is the most abundant gap junction protein in human RPE [11, 17]. We used 10 μM concentration of 18α-GA, which, while only partly blocking gap junctions, does not cause irreversible effects [36]. Membrane capacitance measured in the voltage clamp decreased slightly but statistically significantly, from 29.0 ± 5.5 pF in the presence of nifedipine to 27.3 ± 5.2 pF after application of 18α-GA (n = 6; P = 0.002, paired t-test, Fig. 11F). It should be noted that these measurements were conducted on cells that displayed no delayed inward currents.

In one cell with a delayed inward current, application of 10 μM 18α-GA caused a decrease in Cm and a simultaneous decrease in the delayed current but did not affect the amplitude of the primary INa (Fig. 11G, H). When 18α-GA was washed out and replaced with TTX, Cm returned to the control value while the amplitudes of both inward currents decreased (Fig. 11 I). The results of this experiment summarized in Fig. 11J are consistent with the hypothesis that the delayed inward currents originate in the neighboring cells and are mediated by INa, with signals spreading via gap junction channels.


In this work, we (1) described Na+-based voltage spikes in hESC-RPE cells, (2) documented rapid spontaneous changes in membrane conductance and resting potential, and (3) discovered that electrical excitation can spread to the neighboring cells in a syncytium comprised of the conventionally non-excitable cells.


Expression at significant levels of the classical voltage-activated Na+ conductances characterized by ultra-fast activation and inactivation is usually associated with generation of voltage spiking events, ranging in amplitude from small depolarizing waves to full-blown action potentials. Fast depolarizing voltage responses resembling action potentials have been previously described in the conventionally non-excitable cells such as astrocytes [20,21,22], rabbit pigmented ciliary epithelial cells [23], newt RPE [19], and neonatal rat RPE [16]. However, Botchkin and Matthews attributed the development of INa and the spiking to neuronal trans-differentiation of the cultured and actively proliferating RPE [16].

Previous studies documented expression of multiple types of voltage-activated Na+ and Ca2+ channels both in the cultured hESC-RPE and freshly dissected mouse RPE [4, 5]. INa can be large, raising a question about possible electrogenic function of Na+ channels in RPE. In comparison to the previous study [4], the patch-clamp experiments in this work were performed with series resistance compensated and under conditions as close as possible to those the cells were maintained in, i.e. at the temperature of ~34°C, in the physiological medium and using “all-ions” intracellular solution, so that Vrest was not compromised by added K+ channel blockers. We found that when INa was recovered from inactivation, depolarizing current pulses indeed evoked spikes, the amplitudes of which were mainly determined by the maximal amplitude of INa (Fig. 6A, D). Application of TTX strongly inhibited both INa and spikes (Fig. 4).

We did not detect any spikes triggered spontaneously without prior manipulations with membrane potential. Our results indicate that spontaneous spike generation is precluded by low absolute availability of INa at normal Vrest values and high INa activation threshold when Vrest was permissive. The availability of INa is a function of the maximal INa and the momentary inactivation state. The spikes could not be evoked at the group-average Vrest of −29 ± 18 mV because of near complete INa inactivation. However, in nine cells (Fig. 7F), Vrest transiently hyperpolarized to the levels of <−60 mV, at which a major fraction of INa could recover from inactivation (Fig. 2D–G and 7F–H). Then, if the maximal INa was high enough to yield a sufficiently large residual INa at the Vrest, a depolarizing fluctuation of membrane potential could plausibly trigger a spike as we observed in the current-clamp experiments (Fig. 7). Although we routinely detected spontaneous fluctuations of Vrest, a particularly large and rapid one would be required to reach the INa activation threshold potential at around −40 mV without a fatal loss of INa due to closed-state inactivation. Thus, a combination of several low-probability events is necessary to evoke a distinct spontaneous spike: a large INa expression level, several tens of ms long hyperpolarization to about −70 mV, and a rapid depolarization by about 30 mV (Fig. 7H).

Observations of Vrest fluctuations associated with changes in the putative depolarizing leak conductance (Fig. 8) suggest that addition of even a small repolarizing K+ conductance might be sufficient to permanently lower Vrest. If hESC-RPE cells expressed sufficient quantities of conventional voltage-activated K+ channels [18] supporting sustained current with a low voltage activation threshold, group-average Vrest would be more hyperpolarized. This would not only increase availability of INa but also make the membrane faster, enabling rapid voltage transients. For instance, the cell in Fig. 7G, H is characterized by a very negative Vrest because the depolarizing leak conductance is presumably very small whereas Rm is high. Any transient depolarization can be filtered out by the very slow membrane, preventing rapid reaching of the INa activation threshold.

Interestingly, Vrest of the freshly isolated adult human and monkey RPE was on average −50 mV [12], being in similar range in mouse RPE monolayers [17], that is ~20 mV lower than in our hESC-RPE (differences in liquid junction potentials accounted for). Contributing to Vrest, these cells also had prominent inward and delayed rectifier K+ channels [12] and expressed INa, albeit only in subconfluent cultures [13] similarly to newt RPE [19]. Although the latter observation suggested that INa could be a feature of immature RPE [13], the recent findings demonstrate that Na+ channels in mature hESC- and mouse RPE monolayers are largely localized in the vicinity of the cell-cell junctions [4]. Dissociation of RPE destroys the intercellular contacts and drastically decreases whole-cell INa [4]. Therefore, the observed differences between the freshly isolated and the cultured RPE cells require caution in interpretation of results [4, 13, 16].

Although full-fledged spikes are one of the most spectacular excitation events in hESC-RPE, amplification of small transient depolarizations by the residual INa might be its main electrogenic function. Such depolarizations interspersed with hyperpolarizations are prevalent in our data (Figs. 7A and 9). As discussed previously, depolarizing INa could contribute by promoting the opening of slower L-type Ca2+ channels [37, 38]. Of Ca2+-dependent processes in the RPE, POS phagocytosis is a highly dynamic event requiring rapid actin remodeling for the phagocytic cup formation and phagosome ingestion (reviewed by [39]). As POS phagocytosis has been shown to depend on both L-type Ca2+ and Na+ channels [4, 5, 40, 41], depolarizing INa could facilitate L-type Ca2+ channel opening in the RPE participating in the complex interaction of molecular players involved in the Ca2+ regulation of phagocytosis [2].

Light response decreases the sub-retinal K+ concentration causing hyperpolarization [9] that together with large spontaneous fluctuations in the membrane resistance could trigger activation of INa in vivo and make RPE excitable, with excitation rapidly spreading in the monolayer. In addition, NaV1.8, one of the prominent subtypes present in hESC-RPE [4], has a comparatively depolarized half-inactivation potential of ~ −30 mV [42, 43]. If NaV1.8 contributes to any substantial level, its big portion would be available for activation at rest. Thus, taking into account the presence of several types of voltage-gated ion channels in the RPE, intrinsic properties of RPE membrane and its hyperpolarization in the transition from dark to light, the capacity for spiking makes sense and supports the role of RPE as much more versatile tissue than merely giving homeostatic support to the retina.

Fluctuating membrane leak

Membrane conductance fluctuated in virtually all cells with current-clamp recordings. Similarly, in the voltage clamp experiments, changes in leak occurring in the course of seconds were also observed routinely. The finding of a large, spontaneously and rapidly fluctuating depolarizing conductance is unusual. However, the crucial question is whether the observed changes in Rm reflect genuine, ion channel-dependent changes in membrane conductance, or are due to seal instability. Although this problem cannot be solved decisively without additional pharmacological/genetic studies of ion channels that could mediate leak conductance(s), there are several arguments in support of the physiological leak hypothesis. Changes in leak conductance in the range from 10 to about 0.5 nS localized at the membrane-electrode interface in the whole-cell configuration usually cause irreversible seal disruption. A non-selective instrumental conductance caused by imperfect attachment of plasma membrane to the electrode glass would dissipate the local transmembrane ionic gradients in the vicinity of the membrane-electrode interface. This would immediately diminish the recorded voltage-activated ionic currents as is observed with bad seals, and lead to quick run-down. Our experience with other types of cells suggests that although seal quality can slowly improve over time, especially in the beginning of an experiment, and thus increase whole-cell resistance, the opposite leak-increasing changes quickly ruin the experiment. Continuous and rapid fluctuations of leak occurring repeatedly over the course of many minutes are inconsistent with the seal instability hypothesis.

Assuming that the fluctuating leak was physiological, what channels could mediate it? The leak current is overall depolarizing, indicating that K+ channels in the plasma membrane cannot dominate it above the K+ reversal potential of −90 mV. However, although membrane properties appeared to be passive both in high- and low-resistance states (Figs. 3B, E and 8F) suggesting that leak channels are not voltage-activated, changes in leak appear to be associated with strong changes in membrane potential (Figs. 7A, D and 8G–K). Thus, it is possible that when membrane is strongly hyperpolarized by a current pulse, voltage-dependent inward rectifier K+ channels [44, 45] open and provide rapid depolarization, contributing to the sharp depolarizing transients we observed (Fig. 9). However, these channels tend to close with depolarization (Fig. 1E). At membrane potentials more negative than the reversal potential for Cl (−39 mV in our experiments), efflux of Cl via various chloride channels [6, 8] could mediate a depolarizing leak current.

However, we argue that leak current fluctuations are probably caused by changes in gap junction conductance [46, 47]. Under physiological conditions, RPE cells face three external ionic milieus, two on the apical and basal surfaces and that/those of the neighboring cells. In our experiments, the trans-epithelial chemical gradient was disrupted but the intercellular ionic gradient probably exacerbated after the cytosol equilibrated with the patch pipette solution. RPE cells express Cx43-based gap junctions [11, 17, 48] that have voltage-dependent gating properties [47], so a strong depolarization or hyperpolarization of the clamped cell can change the transjunctional potential, leading to the opening or closing of gap junctions. This is consistent with the observation that sudden changes in leak were associated with strong changes in membrane potential during current-clamp experiments (Fig. 7A, D). Furthermore, recent studies reveal the presence of functional Cx43 hemichannels largely on the apical membrane of human and mouse RPE [11, 17]. Opening or closing of Cx43 hemichannels expressed in the plasma membrane could contribute to the changes in Rm observed in this work.

Spread of electrical excitation in RPE

Delayed activation of sodium currents characterized by more positive reversal potential than that of the normal INa is often observed in neurons with elongated branches. The so-called escaping currents are caused by inadequate space clamp [49, 50]. In the small hexagonally shaped RPE cells, however, INa can escape not from a distal dendrite or axon but from the neighboring cells connected with open gap junctions.

Gap junctional coupling between cells in the RPE monolayer is likely to promote synchronization of various processes, and its role in eye development has been shown in the developing chick [48, 51]. While it is known that gap junctions in RPE allow the spread of electrical signals [17, 52], here we demonstrated evidence that electrical coupling could be strong enough to allow spread of INa-mediated excitation waves to the neighbors of the patched cell (Fig. 10C, F).

We propose the following mechanism for the generation of the delayed inward currents observed in this study. When the patched cell is connected to its neighbors via open gap junctions, they introduce additional junctional resistance (RGJ) in series with the electrode series resistance (Rs), so that charging of the neighboring cell is governed by the sum of two series resistances and its Rm, and the higher the ratio of RGJ + Rs to the second cell’s Rm, the smaller the voltage drop on the membrane. Consequently, the neighboring cells can be clamped via gap junctions but the eventual membrane voltage and the speed of its change will depend on RGJ + Rs. When RGJ is relatively small, the clamping of the second cell, while accomplished with probably tens of mV voltage errors, could still yield enough INa to be detected across the junction. Conversely, when RGJ is large, the other cell’s voltage cannot be commanded to a degree sufficient to either recover INa or reach its activation threshold. Thus, multiple delayed currents probably arise in different neighboring cells with dissimilar but permissive RGJ values, and if RGJ values of two or more connections are about the same, their delayed currents would merge. Similar but comparatively slow processes can occur during current injections in the current-clamp experiments.

It is even possible that if RGJ is very low, providing large conductance, the delayed current can merge with the primary INa. If RGJ can indeed fluctuate dynamically then this might explain the broadening of INa in Fig. 10H right sub-panel compared to the left sub-panel. However, such cells would still be characterized by slow settling of the capacitance transient in the voltage clamp indicative of clamping across the junctions.

It appears that in the absolute majority of RPE cells the RGJ was too high to enable delayed inward currents. This is supported by our experiments with Cx43 inhibitor 18α-GA, which resulted in a significant but comparatively small effect on membrane capacitance (Fig. 11F). While the cause of observed variability in gap junctional conductance is not clear, in addition to the transjunctional potential, gap junctions are regulated by many mechanisms, including internal Ca2+ concentration, elevation of which can rapidly and completely close the gap junction channels [35, 53, 54]. It was reported that increase of the cytosolic Ca2+ concentration after application of the calcium ionophore ionomycin resulted in a reduction of Cx43 gap junction conductance by a chemical gating mechanism [53]. In RPE, calcium can enter the cell from the extracellular space via TRP and L-type Ca2+ channels, from the endoplasmic reticulum via IP3R channels, and from the neighboring cells via gap junctions [6, 51], and these sources might manifest dissimilarly in different cells.

It is possible that in the cells where delayed inward currents were observed, L-type Ca2+ channels could be the main source of Ca2+ influx. When voltage in such cells is held below the L-type Ca2+ channel activation threshold, either by clamping membrane potential to −80 mV in the voltage clamp, or by injecting a hyperpolarizing current in the current clamp, the influx of Ca2+ would be minimized and the gap junction blockade relieved. However, when the cell is depolarized during the test pulses, inward Ca2+ current would resume, partly blocking the gap junction channels. Then, as nifedipine blocked the L-type Ca2+ channels, Ca2+ influx stopped even during the depolarizations, thus maximizing the gap junction conductance, and increasing the delayed inward current amplitude and speeding up its onset (Fig. 11). Alternatively, cells characterized by the delayed inward currents could express more gap junction channels so that the same fraction of gap junction conductance as in the “normal” cells would produce sufficiently low RGJ to elicit INa. Finally, it is worth keeping in mind that Cx43 gap junctions are suggested to be involved in ephaptic coupling, e.g., in vertebrate retina [55] and heart [56], strengthening the spread of depolarization to the nearby cells. We have previously shown that in RPE, both gap junctions and voltage-gated Na+ channels localize to the tight junctions between the cells [4, 17] generating conditions where neighboring cell membranes containing these channels are closely associated. This could support ephaptic coupling eliciting an intriguing aspect for future studies.


The present work reveals how non-neuronal RPE cells residing in vivo in the immediate interactive contact with neuronal retina have active intrinsic properties. Remarkably, these cells can change the properties not only as response to the relatively slow changes in the ionic milieu associated with illumination-dependent changes in photoreceptor functioning, but also rapidly, on the scale of seconds, influence the electrophysiological states of the neighboring cells. Our studies demonstrate that in RPE, currents through Na+ channels mediate voltage spikes that can spread laterally in the monolayer through gap junctions. In addition, membrane resistance in the RPE shows immense and spontaneous fluctuations causing correlated alterations in the membrane potential. These properties enable mutual and previously uncovered interaction between RPE and the neural retina yet requiring further investigation with undisturbed physical interaction between the two tissues.


Cell culture

Two previously derived hESC lines Regea08/017 and Regea11/013 were used in this study. These lines have been established from the blastocyst-stage human embryos using mechanical derivation methods [57]. The hESC lines were cultured and differentiated into RPE as previously described [58, 59] followed by cryopreservation [28, 60]. The cells were plated on hanging cell culture inserts (polyethylene terephthalate membrane with 1.0 μm pore size, Sarstedt, Inc., Newton, NC, USA) coated with Biolaminin LN 521 (1.8 μg/cm2) and Collagen IV (10 μg/cm2) with a density of 5.5 × 105 cells/cm2. Inserts were placed in 24-well plates (Corning Inc., Corning, NY, USA) and cell monolayers nourished from both the apical and basolateral sides. For our experiments, the cells were cultured at +37 °C in 5% CO2 in Knock-Out Dulbecco’s modified Eagle’s medium (KO-DMEM) containing 15% Knock-Out serum replacement (KO-SR), 1% Glutamax, 0.2% 2-mercaptoethanol (all from Thermo Fisher Scientific, Waltham, MA, USA), 1% non-essential amino acids (NEAA), and 50 U/mL penicillin/streptomycin (both from Lonza Group, Basel, Switzerland). The medium was replenished three times a week.

Patch-clamp recordings

Cells cultured for 4–26 weeks from several independent differentiation batches were used. Current and voltage recordings were obtained from hESC-RPE monolayers using the standard patch-clamp technique in the whole-cell configuration. Patch-clamp pipettes were pulled from borosilicate glass capillaries (0.86 mm I.D. 1.5 mm O.D., Science Products GmbH, Hofheim, Germany and Harvard apparatus, Holliston, MA, USA) using a filament-based puller P-1000 (Sutter Instrument, Novato, CA, USA). Pipettes were filled with an internal solution containing (in mM): 83 K-gluconate, 25 KCl, 5.5 EGTA, 0.5 CaCl2, 4 ATP-Mg, 0.1 GTP-Na, 10 HEPES, and 5 NaCl (pH adjusted to 7.2 with KOH and osmolarity to 290 mOsm with sucrose). The final electrode resistance was between 4 and 7 MΩ. In some experiments (Fig. 11F–K and Additional file 1. Fig. S1), K-gluconate (83 mM) and KCl (25 mM) in the pipette solution were replaced with CsCH3SO3 (83 mM) and CsCl (25 mM).

RPE monolayers were continuously superfused with Ames’ solution (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10 mM HEPES and 10 mM NaCl (pH adjusted to 7.4 with NaOH and osmolarity to 315 mOsm with sucrose).

Extracellular solution was heated to 33–35 °C with the inline heater SH-27B controlled with a TC-324B unit (Warner Instruments, Hamden, CT, USA). In some experiments, nifedipine or 18α-glycyrrhetinic acid (both 10 μM, Sigma-Aldrich, St. Louis, MO, USA), or TTX citrate (10 μM, Abcam, Cambridge, UK) was added to the bath solutions.

All recordings were made using Axopatch 200B patch-clamp amplifier connected to Digidata 1440A converter (Molecular Devices, San Jose, CA, USA). Currents were recorded at the sampling frequency of 250 kHz and low-pass filtered at 2 kHz. Holding potential was −70 mV in all voltage clamp experiments. Resting membrane potential (Vrest) was determined from the current clamp in the absence of current injection. Only recordings with series resistance of ≤ 25 MΩ were used. Liquid junction potential calculated according to the Nernst-Planck equation was +13 mV. Data presented in the text and figures were not compensated for the liquid junction potential. Capacitance compensation was used to full extent. In the majority of cells, series resistance was compensated by the built-in circuit of the amplifier (20–80% correction).

Equilibrium potentials calculated from the patch pipette and bath solutions concentrations were −90.1 mV for EK, 85.8 mV for ENa, 221.7 mV for ECa, and −39.3 mV for ECl.

Data analysis and statistics

The character of data distribution was measured using the Shapiro-Wilk normality test. Data in the samples that generally did not pass the normality test are presented in the text as medians with 1st:3rd interquartile ranges. In pairwise comparisons, such data were statistically compared using Mann–Whitney U-test (MWUT). All normally distributed cumulative data were presented as means ± standard deviation (s.d.). Significance testing was performed using the t-test as indicated. Pearson’s correlation coefficient (r) was used for the analyses of correlations. Throughout the text, (n) denotes the group size.

Availability of data and materials

All data generated or analyzed during this study are included in this published article, its supplementary information files (Additional file 1: Fig. S1, Additional file 2: Table S1), and Figshare data repository [61]. Where n < 6, the individual data values are provided in additional files (Additional file 2: Table S1) and cited in the figure legends.


C m :

Membrane capacitance

g :



Human embryonic stem cell-derived retinal pigment epithelium

I C :

Capacitive current

I IR :

Sustained hyperpolarization-activated inward current

I Na :

Voltage-activated Na+ current

I K :

Voltage-activated K+ current

I A :

Transient outward voltage-activated K+ current

I DR :

Sustained voltage-activated K+ current


Photoreceptor outer segments

R m :

Membrane resistance

R GJ :

Junctional resistance

R s :

Series resistance


Steady-state inactivation



V rest :

Resting potential

τ m :

Membrane time constant

τ rec :

Recovery time constant


18α-glycyrrhetinic acid


  1. Sparrow JR, Hicks D, Hamel CP. The retinal pigment epithelium in health and disease. Curr Mol Med. 2010;10(9):802–23.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Strauss O. The retinal pigment epithelium in visual function. Physiol Rev. 2005;85(3):845–81.

    Article  CAS  PubMed  Google Scholar 

  3. Giblin JP, Comes N, Strauss O, Gasull X. Ion channels in the eye: involvement in ocular pathologies. Adv Protein Chem Struct Biol. 2016;104:157–231.

    Article  CAS  PubMed  Google Scholar 

  4. Johansson JK, Karema-Jokinen VI, Hakanen S, Jylhä A, Uusitalo H, Vihinen-Ranta M, et al. Sodium channels enable fast electrical signaling and regulate phagocytosis in the retinal pigment epithelium. BMC Biology. 2019;17(1):63.

    Article  PubMed  PubMed Central  Google Scholar 

  5. Korkka I, Viheriala T, Juuti-Uusitalo K, Uusitalo-Jarvinen H, Skottman H, Hyttinen J, et al. Functional voltage-gated calcium channels are present in human embryonic stem cell-derived retinal pigment epithelium. Stem Cells Transl Med. 2019;8(2):179–93.

    Article  CAS  PubMed  Google Scholar 

  6. Wimmers S, Karl MO, Strauss O. Ion channels in the RPE. Prog Retin Eye Res. 2007;26(3):263–301.

    Article  CAS  PubMed  Google Scholar 

  7. Hughes BA, Takahira M, Segawa Y. An outwardly rectifying K+ current active near resting potential in human retinal pigment epithelial cells. Am J Physiol. 1995;269(1 Pt 1):C179-87.

    Article  CAS  PubMed  Google Scholar 

  8. Reichhart N, Strauss O. Ion channels and transporters of the retinal pigment epithelium. Exp Eye Res. 2014;126:27–37.

    Article  CAS  PubMed  Google Scholar 

  9. Strauss O. Transport mechanisms of the retinal pigment epithelium to maintain of visual function. Heat and Mass Transfer. 2014;50(3):303–13.

    Article  CAS  Google Scholar 

  10. Rosenthal R, Heimann H, Agostini H, Martin G, Hansen LL, Strauss O. Ca2+ channels in retinal pigment epithelial cells regulate vascular endothelial growth factor secretion rates in health and disease. Mol Vis. 2007;13:443–56.

    CAS  PubMed  PubMed Central  Google Scholar 

  11. Akanuma SI, Higashi H, Maruyama S, Murakami K, Tachikawa M, Kubo Y, et al. Expression and function of connexin 43 protein in mouse and human retinal pigment epithelial cells as hemichannels and gap junction proteins. Exp Eye Res. 2018;168:128–37.

    Article  CAS  PubMed  Google Scholar 

  12. Wen R, Lui GM, Steinberg RH. Whole-cell K+ currents in fresh and cultured cells of the human and monkey retinal pigment epithelium. J Physiol. 1993;465:121–47.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  13. Wen R, Lui GM, Steinberg RH. Expression of a tetrodotoxin-sensitive Na+ current in cultured human retinal pigment epithelial cells. J Physiol. 1994;476(2):187–96.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Sakai H, Saito T. Na+ and Ca2+ channel expression in cultured newt retinal pigment epithelial cells: comparison with neuronal types of ion channels. J Neurobiol. 1997;32(4):377–90.

    Article  CAS  PubMed  Google Scholar 

  15. Pattnaik BR, Hughes BA. Effects of KCNQ channel modulators on the M-type potassium current in primate retinal pigment epithelium. Am J Physiol Cell Physiol. 2012;302(5):C821-33.

    Article  CAS  PubMed  Google Scholar 

  16. Botchkin LM, Matthews G. Voltage-dependent sodium channels develop in rat retinal pigment epithelium cells in culture. Proc Natl Acad Sci U S A. 1994;91(10):4564–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Fadjukov J, Wienbar S, Hakanen S, Aho V, Vihinen-Ranta M, Ihalainen TO, et al. Gap junctions and connexin hemichannels both contribute to the electrical properties of retinal pigment epithelium. J Gen Physiol. 2022;154(4):202112916.

    Article  Google Scholar 

  18. Korkka I, Skottman H, Nymark S. Heterogeneity of potassium channels in human embryonic stem cell-derived retinal pigment epithelium. Stem Cells Transl Med. 2022;11(7):753–66.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Sakai H, Saito T. Development of voltage-dependent inward currents in dissociated newt retinal pigment epithelial cells in culture. Neuroreport. 1994;5(8):933–6.

    Article  CAS  PubMed  Google Scholar 

  20. Barres BA, Chun LL, Corey DP. Ion channel expression by white matter glia: I. type 2 astrocytes and oligodendrocytes. Glia. 1988;1(1):10–30.

    Article  CAS  PubMed  Google Scholar 

  21. Sontheimer H, Black JA, Ransom BR, Waxman SG. Ion channels in spinal cord astrocytes in vitro. I. transient expression of high levels of Na+ and K+ channels. J Neurophysiol. 1992;68(4):985–1000.

    Article  CAS  PubMed  Google Scholar 

  22. Sontheimer H, Waxman SG. Ion channels in spinal cord astrocytes in vitro. II. biophysical and pharmacological analysis of two Na+ current types. J Neurophysiol. 1992;68(4):1001–11.

    Article  CAS  PubMed  Google Scholar 

  23. Fain GL, Farahbakhsh NA. Voltage-activated currents recorded from rabbit pigmented ciliary body epithelial cells in culture. J Physiol. 1989;418:83–103.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Forest DL, Johnson LV, Clegg DO. Cellular models and therapies for age-related macular degeneration. Dis Model Mech. 2015;8(5):421–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  25. Schwartz SD, Regillo CD, Lam BL, Eliott D, Rosenfeld PJ, Gregori NZ, et al. Human embryonic stem cell-derived retinal pigment epithelium in patients with age-related macular degeneration and Stargardt’s macular dystrophy: follow-up of two open-label phase 1/2 studies. Lancet. 2015;385(9967):509–16.

    Article  PubMed  Google Scholar 

  26. Sharma R, Bose D, Maminishkis A, Bharti K. Retinal pigment epithelium replacement therapy for age-related macular degeneration: are we there yet? Annu Rev Pharmacol Toxicol. 2020;60:553–72.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Limnios IJ, Chau YQ, Skabo SJ, Surrao DC, O’Neill HC. Efficient differentiation of human embryonic stem cells to retinal pigment epithelium under defined conditions. Stem Cell Res Ther. 2021;12(1):248.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Hongisto H, Jylha A, Nattinen J, Rieck J, Ilmarinen T, Vereb Z, et al. Comparative proteomic analysis of human embryonic stem cell-derived and primary human retinal pigment epithelium. Sci Rep. 2017;7(1):6016.

    Article  PubMed  PubMed Central  Google Scholar 

  29. Carr AJ, Smart MJ, Ramsden CM, Powner MB, da Cruz L, Coffey PJ. Development of human embryonic stem cell therapies for age-related macular degeneration. Trends Neurosci. 2013;36(7):385–95.

    Article  CAS  PubMed  Google Scholar 

  30. Sorkio A, Hongisto H, Kaarniranta K, Uusitalo H, Juuti-Uusitalo K, Skottman H. Structure and barrier properties of human embryonic stem cell-derived retinal pigment epithelial cells are affected by extracellular matrix protein coating. Tissue Eng Part A. 2014;20(3–4):622–34.

    CAS  PubMed  PubMed Central  Google Scholar 

  31. Burgess DE, Crawford O, Delisle BP, Satin J. Mechanism of inactivation gating of human T-type (low-voltage activated) calcium channels. Biophys J. 2002;82(4):1894–906.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. Hille B. Ion Channels of Excitable Membranes. 3rd ed. Sunderland, MA: Sinauer Associates; 2001.

    Google Scholar 

  33. Goldman L. Internal cesium and the sodium inactivation gate in Myxicola giant axons. Biophys J. 1986;50(2):231–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Ignatova II, French AS, Torkkeli PH, Liu H, Frolov RV. Suppression of Gq and PLC gene expression has a small effect on quantum bumps in vivo in Periplaneta americana. J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 2020;206(4):597–610.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Peracchia C. Calmodulin-mediated regulation of gap junction channels. Int J Mol Sci. 2020;21(2):485.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Guo Y, Martinez-Williams C, Gilbert KA, Rannels DE. Inhibition of gap junction communication in alveolar epithelial cells by 18alpha-glycyrrhetinic acid. Am J Physiol. 1999;276(6):L1018-26.

    CAS  PubMed  Google Scholar 

  37. Roden DM, Balser JR, George AL Jr, Anderson ME. Cardiac ion channels. Annu Rev Physiol. 2002;64:431–75.

    Article  CAS  PubMed  Google Scholar 

  38. Prajapati C, Polonen RP, Aalto-Setala K. Simultaneous recordings of action potentials and calcium transients from human induced pluripotent stem cell derived cardiomyocytes. Biol Open. 2018;7(7):035030.

    Google Scholar 

  39. Lakkaraju A, Umapathy A, Tan LX, Daniele L, Philp NJ, Boesze-Battaglia K, et al. The cell biology of the retinal pigment epithelium. Prog Retin Eye Res. 2020:100846.

  40. Karl MO, Kroeger W, Wimmers S, Milenkovic VM, Valtink M, Engelmann K, et al. Endogenous Gas6 and Ca2+ -channel activation modulate phagocytosis by retinal pigment epithelium. Cell Signal. 2008;20(6):1159–68.

    Article  CAS  PubMed  Google Scholar 

  41. Muller C, Mas Gomez N, Ruth P, Strauss O. CaV1.3 L-type channels, maxiK Ca(2+)-dependent K(+) channels and bestrophin-1 regulate rhythmic photoreceptor outer segment phagocytosis by retinal pigment epithelial cells. Cell Signal. 2014;26(5):968–78.

    Article  PubMed  Google Scholar 

  42. Akopian AN, Sivilotti L, Wood JN. A tetrodotoxin-resistant voltage-gated sodium channel expressed by sensory neurons. Nature. 1996;379(6562):257–62.

    Article  CAS  PubMed  Google Scholar 

  43. Han C, Vasylyev D, Macala LJ, Gerrits MM, Hoeijmakers JG, Bekelaar KJ, et al. The G1662S NaV1.8 mutation in small fibre neuropathy: impaired inactivation underlying DRG neuron hyperexcitability. J Neurol Neurosurg Psychiatry. 2014;85(5):499–505.

    Article  PubMed  Google Scholar 

  44. Huang H, Li H, Shi K, Wang L, Zhang X, Zhu X. TREKTRAAK twopore domain potassium channels protect human retinal pigment epithelium cells from oxidative stress. Int J Mol Med. 2018;42(5):2584–94.

    CAS  PubMed  PubMed Central  Google Scholar 

  45. Shen C, Ma W, Zheng W, Huang H, Xia R, Li C, et al. The antioxidant effects of riluzole on the APRE-19 celll model injury-induced by t-BHP. BMC Ophthalmol. 2017;17(1):210.

    Article  PubMed  PubMed Central  Google Scholar 

  46. Bukauskas FF, Verselis VK. Gap junction channel gating. Biochim Biophys Acta. 2004;1662(1–2):42–60.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  47. Gonzalez D, Gomez-Hernandez JM, Barrio LC. Molecular basis of voltage dependence of connexin channels: an integrative appraisal. Prog Biophys Mol Biol. 2007;94(1–2):66–106.

    Article  CAS  PubMed  Google Scholar 

  48. Pearson RA, Dale N, Llaudet E, Mobbs P. ATP released via gap junction hemichannels from the pigment epithelium regulates neural retinal progenitor proliferation. Neuron. 2005;46(5):731–44.

    Article  CAS  PubMed  Google Scholar 

  49. Sontheimer H, Ransom CB. Whole-cell patch-clamp recordings. In: Walz W, Boulton AA, Baker GB, editors. Patch-clamp analysis: advanced techniques. Totowa, NJ: Humana Press; 2002. p. 35–67.

    Chapter  Google Scholar 

  50. Milescu LS, Bean BP, Smith JC. Isolation of somatic Na+ currents by selective inactivation of axonal channels with a voltage prepulse. J Neurosci. 2010;30(22):7740–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  51. Pearson RA, Catsicas M, Becker DL, Bayley P, Luneborg NL, Mobbs P. Ca(2+) signalling and gap junction coupling within and between pigment epithelium and neural retina in the developing chick. Eur J Neurosci. 2004;19(9):2435–45.

    Article  PubMed  Google Scholar 

  52. Hudspeth AJ, Yee AG. The intercellular junctional complexes of retinal pigment epithelia. Invest Ophthalmol. 1973;12(5):354–65.

    CAS  PubMed  Google Scholar 

  53. Xu Q, Kopp RF, Chen Y, Yang JJ, Roe MW, Veenstra RD. Gating of connexin 43 gap junctions by a cytoplasmic loop calmodulin binding domain. Am J Physiol Cell Physiol. 2012;302(10):C1548-56.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Bennett BC, Purdy MD, Baker KA, Acharya C, McIntire WE, Stevens RC, et al. An electrostatic mechanism for Ca(2+)-mediated regulation of gap junction channels. Nat Commun. 2016;7:8770.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  55. Vroman R, Klaassen LJ, Kamermans M. Ephaptic communication in the vertebrate retina. Front Hum Neurosci. 2013;7:612.

    Article  PubMed  PubMed Central  Google Scholar 

  56. Veeraraghavan R, Hoeker GS, Alvarez-Laviada A, Hoagland D, Wan X, King DR, et al. The adhesion function of the sodium channel beta subunit (β1) contributes to cardiac action potential propagation. eLife. 2018;7:e37610.

    Article  PubMed  PubMed Central  Google Scholar 

  57. Skottman H. Derivation and characterization of three new human embryonic stem cell lines in Finland. In Vitro Cell Dev Biol Anim. 2010;46(3–4):206–9.

    Article  PubMed  Google Scholar 

  58. Vaajasaari H, Ilmarinen T, Juuti-Uusitalo K, Rajala K, Onnela N, Narkilahti S, et al. Toward the defined and xeno-free differentiation of functional human pluripotent stem cell-derived retinal pigment epithelial cells. Mol Vis. 2011;17:558–75.

    CAS  PubMed  PubMed Central  Google Scholar 

  59. Skottman H, Muranen J, Lahdekorpi H, Pajula E, Makela K, Koivusalo L, et al. Contacting co-culture of human retinal microvascular endothelial cells alters barrier function of human embryonic stem cell derived retinal pigment epithelial cells. Exp Cell Res. 2017;359(1):101–11.

    Article  CAS  PubMed  Google Scholar 

  60. Hongisto H, Ilmarinen T, Vattulainen M, Mikhailova A, Skottman H. Xeno- and feeder-free differentiation of human pluripotent stem cells to two distinct ocular epithelial cell types using simple modifications of one method. Stem Cell Res Ther. 2017;8(1):291.

    Article  PubMed  PubMed Central  Google Scholar 

  61. Ignatova I, Frolov R, Nymark S. Electrical excitability and its lateral spread in the retinal pigment epithelium: dataset. 2023.

Download references


We acknowledge Sanna Yrjänheikki, Outi Melin, and Hanna Pekkanen for technical assistance and Tampere Facility of Electrophysiological Measurements for their services. Dr. Heli Skottman is acknowledged for providing her expertise on hESC-RPE, and Dr. Gregory Schwartz for valuable suggestions on the manuscript.


Open access funding provided by Tampere University including Tampere University Hospital, Tampere University of Applied Sciences (TUNI). This research was financially supported by the Academy of Finland (to SN, grant numbers 319257 and 323507), by Mary and Georg Ehrnrooth foundation (to II, grant number 201900005) and by the Silmä- ja kudospankkisäätiö (to II, grant number 201900001).

Author information

Authors and Affiliations



Conception and design of the study was done by all authors. Patch-clamp experiments were performed by II, data analysis by II and RF, and interpretation of the data and writing of the manuscript by all authors. The authors read and approved the final manuscript.

Author’s information

Twitter handle: @nymarklab (Soile Nymark)

Corresponding author

Correspondence to Soile Nymark.

Ethics declarations

Ethics approval and consent to participate

Tampere University has approval (Dnro 1426/32/300/05) from the National Supervisory Authority for Welfare and Health, Finland, for the research with human embryos. The local Ethical Committee of the Pirkanmaa Hospital District, Finland, has granted a supportive statement to derive, expand, and differentiate hESC lines from surplus embryos for research purposes (R05116). No new cell lines have been derived in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1:

Fig. S1. Kinetics of INa recorded using cesium-based intracellular solution

Additional file 2:

Table S1. Datasets

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ignatova, I., Frolov, R. & Nymark, S. The retinal pigment epithelium displays electrical excitability and lateral signal spreading. BMC Biol 21, 84 (2023).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: