Visualization of Dynamic Sub-microsecond Changes in Membrane Potential

Abstract

Direct observation of rapid membrane potential changes is critical to understand how complex neurological systems function. This knowledge is especially important when stimulation is achieved through an external stimulus meant to mimic a naturally occurring process. To enable exploration of this dynamic space, we developed an all-optical method for observing rapid changes in membrane potential at temporal resolutions of ∼25 ns. By applying a single 600-ns electric pulse, we observed sub-microsecond, continuous membrane charging and discharging dynamics. Close agreement between the acquired results and an analytical membrane-charging model validates the utility of this technique. This tool will deepen our understanding of the role of membrane potential dynamics in the regulation of many biological and chemical processes within living systems.

Introduction

Activation of channels and initiation of chemical signaling pathways is intimately coupled to the electric potential across the plasma membrane of living cells. Although the membrane potential is highly regulated in cells to maintain homeostasis, its alteration may indicate a cellular response to various conditions, including changes in the environment, drug application, or cell growth and repair. Patch-clamp electrophysiology has emerged as the standard for measuring membrane potential in cells, leading to an understanding of the interconnection between membrane potential as a key regulator of ion channels and various membrane chemical messengers (1). However, despite its widespread utility, patch-clamp electrophysiology remains limited in recording speed (microseconds) and spatial resolution (whole cell or small sections of membrane in a single experiment). Because this technique requires direct attachment to the plasma membrane via a glass micropipette, the ability to multiplex beyond a few electrodes or obtain measurements of small cellular structures (e.g., dendrites) is not easily overcome (2, 3).

Optical approaches have enabled monitoring of spatial and temporal changes in membrane potential at the cell and whole tissue levels, typically through the addition of an organic fluorescent probe to the membrane (4). Tracking membrane potential at millisecond time resolution is achieved with camera-based microscopy approaches (5) and has been used to measure action potential firing in living brains to map neuron connectivity (6, 7). Sub-microsecond imaging of membrane potential transients, typically to observe transmembrane voltage response to applied electric fields, has been achieved using pulsed laser excitation (8, 9, 10, 11). However, although these techniques provide spatial information during the applied pulse, the single-shot nature of these measurements require the event to be highly repeatable shot to shot to obtain a continuous temporal evolution of the membrane potential dynamics. Time-course measurements of membrane potential from 100-μs electrical pulses have also been obtained; however, these measurements were performed at a single location on a hemispherical bilayer and required significant averaging (12).

To address these limitations, we present a, to our knowledge, new technique for imaging dynamic fluctuations in membrane potential with a temporal resolution of ∼25 ns. By combining an organic fluorescent reporter of membrane potential (Fluovolt; Thermo Fisher Scientific, Waltham, MA) (13) with streak camera imaging, we can resolve the temporal and spatial evolution of membrane potential during the application of a pulsed electric field. The result is a full profile of membrane potential dynamics in a single shot, allowing for observation of final membrane-charging potential as well as membrane charging and discharging dynamics.

Materials and Methods

Streak imaging

To achieve sub-microsecond time resolution for fluorescence measurements, we employed a Hamamatsu Photonics C7700 streak camera (Hamamatsu City, Japan) imaged onto a Hamamatsu Orca-Flash4.0 sCMOS camera. In short, the streak camera produces a temporal sweep of one-dimensional spatial information. Light passes through an entrance slit and is converted to a proportional number of electrons, which are vertically deflected onto a phosphor screen based on time. To measure rapid changes in membrane voltage, we modified a Nikon (Tokyo, Japan) inverted microscope to allow for epifluorescence laser excitation by a large-frame argon-ion laser tuned to 488 nm (INNOVA 90; Coherent, Santa Clara, CA). The laser was routed through an Uniblitz 6-mm laser shutter (Rochester, NY) to allow for continuous excitation of the sample while reducing phototoxicity by limiting the exposure to just before (1 ms, limited by jitter of shutter) and during the acquisition. Before being coupled into the microscope, the laser beam was expanded by a telescope. Epifluorescence excitation optics remained in the microscope to reimage the expanded laser beam to produce a 50-μm illumination spot at the sample plane through the 60× objective. Laser excitation was directed to the sample and filtered from the fluorescence emission by a long-pass filter set for 488-nm laser excitation (Semrock, Rochester, NY). Irradiance at the sample was ∼1000 W/cm². This spot allowed for even illumination of the entire cell. The fluorescence emission from the camera port of the microscope was expanded by a second telescope and reimaged to the entrance slit of the streak camera to record a single cell response over a large field of view on the streak camera.

Electrical stimulation

Electrical stimulation was delivered to the cell using a custom bipolar electrode consisting of two 125-μm tungsten electrodes placed 110 μm apart. This exposure configuration was described in detail in a previous publication (14). A custom pulse generator was constructed to generate single unipolar and bipolar square wave pulses using a full-bridge voltage-source inverter consisting of four metal-oxide semiconductor field-effect transistors (IXFB38N100Q2, 1 kV; IXYS Corp, Milpitas, CA). Timing was controlled by a digital delay generator (DG535; Stanford Research Systems, Sunnyvale, CA), which generated two control pulses to trigger the digital switches sequentially. Single-pulse exposures were delivered to cells via the bipolar electrode by positioning it 50 μm above the coverslip surface using a stage-mounted micromanipulator (MP285; Sutter Instrument, Novato, CA). The 600-ns pulses with amplitudes of 8–150 V from the power supply were delivered; delivered amplitude was monitored using an oscilloscope (5.5–93 V). Finite difference time domain (FDTD) modeling was used to predict the electric-field amplitude at the cell (200–3460 V/cm). Delivery of the pulses, timing of the laser shutter, and triggering of the streak camera were controlled by a Stanford DG645 Delay Generator (Sunnyvale, CA). Electrical pulses were timed to trigger at 1 μs during the 5-μs sweep, based on the manufacturer’s calibrated sweep delays.

Cell culture and fluorophore loading

Chinese hamster ovarian-K1 (CHO-K1) cell lines were obtained from American Type Culture Collection (ATCC, Manassas, VA). CHO-K1 cells were propagated at 37^oC with 5% CO₂ in air, in F-12K medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and 100 IU/mL penicillin/streptomycin (ATCC). To simplify the image collection and analysis, spherical cells were ideal. After replating, CHO-K1 cells maintain a nearly spherical geometry for up to 4 h despite adherence to the coverslip. Four hours before imaging, the cells were trypsinized, according to ATCC protocol, and plated onto a poly-D-lysine 35-mm cell dish with a glass coverslip bottom (MatTek, Ashland, MA). The cells were then incubated for 3.5 h in an incubator to promote adherence and recovery. Thirty minutes before imaging, cells were loaded with Fluovolt or CellTracker Green CMFDA (Thermo Fisher Scientific, Waltham, MA) in the “exposure buffer”, following the manufacturer’s protocol. “Exposure buffer” was prepared in advance and contained 135 mM NaCl, 5 mM KCl, 10 mM HEPES, 10 mM glucose, 2 mM CaCl₂, and 2 mM MgCl₂ with a pH of 7.4 and osmolality of 290–310 mOsm/kg (all from Sigma-Aldrich, St. Louis, MO). Cells were then placed back into the incubator for 30 min to allow time for the dye to load properly. Just before imaging, the cells were rinsed with fresh exposure buffer.

Image analysis

Streak camera kymographs were opened as a stack in the Fiji distribution of ImageJ, and the pixels corresponding the edges of the fluorescence image were manually determined, without regard for exposure parameter (15). A MATLAB (The MathWorks, Natick, MA) script was written to use these coordinates to obtain the average intensity for the outer 20% of each side of the fluorescence kymograph for each line in the sweep. This analysis generated the temporal response for both the cathode- and anode-facing side of the cell. The background signal was determined by averaging the pixel counts for several cell-width areas of the streak images that contained no fluorescence signal and subtracted from the response. The temporal fluorescence intensity was found to have a systemic variation that was present and of an equivalent pattern in both the sham and CellTracker Green CMFDA images (Fig. 2, a and b). This signal was averaged for all sham exposures and smoothed with a Savitzky-Golay filter. All exposures were divided by this normalized system response. Initial fluorescence intensity, F_o, was determined by averaging the first 150 lines (0.8 μs) before the pulse. The percentage of fluorescence change was calculated as (F − F_o)/F_o. The individual time responses for each exposure parameter (n = 14–20) were then averaged to generate the average temporal response. Data are presented in time from the beginning of the sweep; the pulse was triggered to fire 1 μs into the sweep. Adjustments for electrical and optical delays were made in the final data analysis.

Figure 2.

Temporal signals for applied pulses. (a) Kymographs of Fluovolt for unipolar and bipolar electric pulse exposure with overlaid oscilloscope traces of the pulse. (b) Kymographs of CellTracker Green CMFDA unipolar and bipolar electric pulse exposure with overlaid oscilloscope traces of the pulse. The fluorescence from CellTracker Green CMFDA does not respond to the electric pulse. For (a) and (b), no electric pulse was applied for sham exposures. Each kymograph was smoothed using a Gaussian blur filter of radius 5 (equivalent to 25 ns in the Y dimension and 400 nm in the X dimension) in Fiji and was scaled to the maximal-minimal number of counts. The width of each kymograph is ∼25 μm (c) The average (n = 14) anode and cathode signals (no Gaussian blur filter applied) from the regions of interest within the kymographs from cells exposed to 51 V applied unipolar pulses (blue lines). Oscilloscope traces of the pulses captured during the exposure (red lines). To see this figure in color, go online.

Membrane-charging model

When cells are exposed to an external electric field, a transmembrane voltage, ΔΦ_m, is induced. By solving Laplace’s equation for spherical cells, ΔΦ_m from a homogeneous steady-state electric field is given by Schwan’s equation (16),

$$\mathit{\Delta }{\mathit{\Psi }}_m=3/2ERcos\theta ,$$ (1)

where E is the electric field, R is the cell radius, and θ is the angle from the poles of the cell membrane perpendicular to the electric field. This equation can be used to determine the maximal, steady-state value of the induced transmembrane voltage for direct current field of hundreds of microseconds or more. For more transient behavior, such as alternating current fields with frequencies below 1 MHz or rectangular pulses longer than 1 μs, the first-order Schwan’s equation,

$$\mathit{\Delta }{\mathit{\Psi }}_m=3/2ERcos\theta \left(1-e^{-t/{\tau }_m}\right),$$ (2)

$${\tau }_m=\frac{R{\mathrm{\epsilon }}_m/d_m}{\frac{2{\lambda }_o{\lambda }_i}{2{\lambda }_o+{\lambda }_i}+\frac{R}{d_m}{\lambda }_m}$$ (3)

applies (17, 18), where τ_m is the time constant for membrane charging and is dependent on the membrane thickness (d_m), cell radius (R) and dielectric properties of the cytoplasm (λ_i), cell membrane (λ_m, ε_m), and extracellular medium (λ_o). For higher frequencies or shorter pulses, a second-order extension of Schwan’s equation must be used. These equations have been derived for alternating current fields (19, 20, 21) and for rectangular, trapezoidal, and exponential pulse shapes (20). To our knowledge, these second-order equations have not been experimentally validated because of the brevity of the pulses. However, a number of experimental studies using fluorescent potentiometric dyes, such as di-4-ANEPPS and di-8-ANEPPS, have shown a strong correlation, including the spatial variation, between Schwan’s equation and experimentally obtained fluorescence signal (10, 22). Likewise, the optically detected temporal response of membrane potential has been related to the first-order Schwan’s equation for model membrane systems (12).

Given the brevity of our pulses, we used the second-order extension of Schwan’s equation derived by Kotnik et al. (20) for determining the time course of the transmembrane potential for time-varying electric fields (specifically Appendix A.3, Triangular and Trapezoidal Pulses in that publication). The oscilloscope traces for the applied electric fields were best fit to a trapezoidal function, producing a pulse with linear rise and fall times of 100 ns and a flat top of 525 ns for all voltages applied (Fig. 3 a). Trapezoidal representation of the pulse was used as an approximation of the convolution between the rise time of the pulse and the charging time of the plasma membrane. Membrane charging was modeled in MATLAB (The MathWorks) and determined by the superposition of membrane charging for four-unit ramp functions delayed by the rise and on times of the pulse to produce the trapezoidal shape. Membrane-charging amplitudes were then scaled for each electric field as determined by the oscilloscope trace and FDTD model. To obtain a measurement of cell radius for the model, the μm/pixel for the streak camera kymographs were determined by imaging a United States Air Force resolution target. Average cell radius (7.2 ± 1.6 μm, n = 143) was calculated from the coordinates for the cell edges found in the imaging analysis. Outside solution conductivity was measured to be 1.45 S/m by a SevenMulti conductivity meter (Mettler Toledo, Columbus, OH). Other physiological parameters were taken from literature and are provided in Table 1 (23).

Figure 3.

Model of membrane charging. (a) Delivered signal (blue) was measured by the oscilloscope and superimposed on the trapezoidal simulation pulse (red). (b) Predicted charging potential on the cell for increasing pulse amplitudes. (c) Predicted fluorescence change for cathodic and anodic poles of the cell. (d) Predicted charging potential on the cell for increasing pulse amplitudes for a predicted electric field with 50% efficiency was applied to the cells. (e) Predicted fluorescence change for a predicted electric field with 50% efficiency was applied to the cells. (f) Maximal observed fluorescence change versus predicted membrane charging for 50% applied field amplitude (blue dots) closely matches the Fluovolt calibration curve obtained from patch-clamp electrophysiology. To see this figure in color, go online.

Table 1.

Values of Electric and Dimensional Parameters Used in the Model 23

Parameter	Denotation	Value
Cell radius	R	7.2 μm
Membrane thickness	dm	5 nm
Cytoplasmic conductivity	λi	3.0 × 10−1 S/m
Cytoplasmic permittivity	εi	6.4 × 10−10 As/VmR
Membrane conductivity	λi	3.0 × 10−7 S/m
Membrane permittivity	εi	4.4 × 10−11 As/Vm
Extracellular medium conductivity	λi	1.45 S/m
Extracellular medium permittivity	εi	6.4 × 10−10 As/Vm

Results and Discussion

Acquisition of kymographs of membrane charging

To demonstrate this approach, we applied 600-ns-duration electrical pulses to individual cells observed on an inverted microscope setup (Fig. 1 a). The fluorescence emission image from Fluovolt-loaded CHO-K1 cells was reimaged to the entrance slit of the streak camera to obtain the fluorescence response of the membrane section of the cell perpendicular to the electrodes (Fig. 1 b) This configuration generated a kymograph of the fluorescence intensity of this section of the cell in time versus distance across the cell. CHO-K1 cells, which lack voltage-gated channels, were chosen for these experiments to limit the effect of the cell response to the applied field on the membrane potential. Single 600-ns electrical pulses were applied to test the dynamic response of the streak system. Applied voltages ranged from 5.5 to 93 V, resulting in predicted electric-field amplitudes of 0.2–3.5 kV/cm at the cell, as determined by FDTD modeling (Fig. 1 d) (14). Although the FDTD model predicts a peak electric field, 80% of this value is used to account for the small area of the peak field and obtain an average field across the cells within the exposure region. For this study, nanosecond-duration pulses were applied at modest voltages, thereby enabling the rapid charging of the plasma membrane while reducing the likelihood of severe membrane disruption known to be caused by longer- or higher-amplitude pulses (24, 25).

Figure 1.

Streak microscopy system for high-speed acquisition of membrane potential in a living cell. (a) A schematic of the microscope system contains the streak camera and pulsing system. (b) The electrodes are aligned perpendicular to the entrance slit of the streak camera image to produce the greatest change in membrane potential at the poles of the cell (bottom). A Fluovolt fluorescence image from a single cell was captured by the streak camera with the slit open; the approximate width and typical position of the slit for streak acquisition is shown in gray (top). (c) Kymograph of the equatorial cell slice is for an electric pulse exposure; the oscilloscope trace for the exposure is superimposed on the kymograph. (d) Finite difference time domain (FDTD) model results for the electric field at the cell plane. Model results are given in V/m for 1 V applied and scaled to the applied field. Voltage supplied was delivered by the power supply to the pulse generator. Applied voltage was the potential delivered to the electrodes as measured by an oscilloscope across a parallel resistor. FDTD field is the peak electric field for the applied voltage. The 80% value is used as an average field felt by the cells to avoid nonuniformity issues between the electrodes and a very small peak region. To see this figure in color, go online.

Kymographs from the streak camera demonstrate higher fluorescence signals near the outside of the cells, indicative of membrane loading of the Fluovolt dye (Fig. 2 a). Application of the single 600-ns electrical pulse caused a sudden increase in signal on the side of the cell nearest the cathode electrode, indicative of depolarization of the membrane, along with a corresponding decrease in signal near the anode electrode, indicative of hyperpolarization. To determine reversibility of this fluorescence signal, we applied a bipolar electric pulse, consisting of a 300-ns positive polarity pulse immediately followed by a 300-ns negative polarity pulse. We observed brief depolarization and hyperpolarization upon exposure with a corresponding reversal of the signal with polarity shift of the applied electric field. To ensure that the signals were not an artifact of the electrical pulse, CHO-K1 cells were loaded with CellTracker Green CMFDA Dye, a cytoplasmic-loading fluorescence dye, which should not be affected by membrane potential. The electrical pulse had no effect on the streak fluorescence signal from CellTracker Green CMFDA Dye (Fig. 2 b).

Quantification of membrane charging

To quantify the fluorescence change observed from the nanosecond electrical pulse exposure, the outer edges of the fluorescence signal from each cell were determined manually and independent of exposure labels. The time-varying fluorescence signal was then determined by taking the average intensity for each line in the kymograph across the outer 20% of each side of the fluorescence streak. Temporal traces of the average cathode and anode signal from 14 cells are superimposed on the applied voltage (51 V) trace from an oscilloscope (Fig. 2 c). A correlation between the applied signals and the change in membrane potential on the sides of the cell was observed. For these exposures, the fluorescence signal climbed relatively slowly (in relation to the pulse duration) to a plateau near the end of the applied pulse. Additionally, despite the rather high voltage applied, the signal returned to baseline at a simliar rate after the pulse was removed, which suggests that both Fluovolt and the plasma membrane were not permanently altered within this time frame.

The strength of the membrane charging depends on a cosine relationship to the orientation of the applied electric field (26). To validate the role of membrane charging in the change in the Fluovolt intensity, we rotated the stimulation electrodes parallel to the streak camera slit. As expected, this rotation resulted in only minimal change in the Fluovolt signal across the kymograph (Fig. S1).

Modeling the membrane charging and calibrating the fluorescence response

As shown in Fig. 3 b, the membrane-charging model resulted in a membrane-charging potential approaching 2 V, for exposures up to 51 V applied, in which the maximal potential is reached at the end of the electric pulse. As shown in this model, the membrane-charging potential does not reach a steady-state value during the duration of these pulses. This result matches previous modeling results, which indicated that rectangular pulses shorter than 1 μs would not induce steady-state voltages predicted by ΔΨ_m = 3/2ERcosθ (20). Likewise, the membrane-discharging time constant for these exposures also approaches 1 μs.

Using a voltage-step protocol, the percentage of fluorescence change in relation to membrane potential was found to be nearly linear from −100 to 80 mV for Fluovolt within CHO-K1 cells (Fig. S2, a and b). To determine the extent of the Fluovolt response beyond the typical physiological regime, we also monitored the fluorescence signal and current as we attempted to hold voltages beyond ±200 mV. We found that although the cells tolerated negative potentials, potentials of +200 mV resulted in significant current leak after tens of milliseconds (Fig. S2 d). Thus, for these measurements, we changed the protocol to jump immediately to higher potentials (Fig. S2 e). For these larger voltage jumps, we observed a nonlinear and unstable (with the exception of −200 mV) response of the fluorescence signal with change in membrane potential (Fig. S2, f and g).

Given the unstable nature of the Fluovolt signal at severely high and low membrane potentials, the fluorescence change for membrane potentials for the streak camera experiments beyond −0.5 and +0.5 V were set to the maximal fluorescence change obtainable from the system, determined by applying voltages of 93 V to individual cells. The results from the voltage-step protocol (−100 to 80 mV), along with the saturation points, were then fit to a four-parameter sigmoidal curve in GraphPad Prism 7. As CHO-K1 cells have a resting membrane potential of ∼−60 mV, the fluorescence calibration curve was shifted so that a potential of −60 mV was equivalent to a membrane-charging potential of 0 V. The sigmoidal response of the Fluovolt dye was used to convert the predicted membrane charging to a percentage of fluorescence change (Fig. 3 c).

Comparison of experimental and model results

We used the predicted time-varying fluorescence signal to determine the ability of our system to accurately map the dynamic change in membrane potential through the application of an electric field. Of particular interest was the membrane charging and discharging during the applied electric field. Raw single-shot streak images were found to have a rather high-voltage-equivalent noise signal of 80 mV, so a Gaussian blur filter of radius five pixels (equivalent to 25 ns in the time dimension) was applied to the individual streak images in Fiji (15) before processing to reduce the noise signal for individual acquisitions to 35 mV. Averaging the response from 10 or more, these filtered acquisitions further reduced the noise to <13 mV. The calibrated predicted fluorescence changes for the predicted average electric field (80% of maximum) were superimposed on the measured change in fluorescence for four electric pulse amplitudes (Fig. 4, blue traces). Averages from the raw (unfiltered) streak acquisitions are provided in Fig. S3.

Figure 4.

Signals measured from the streak microscope compared with membrane-charging model predictions. (a–d) Raw data (n = 10–14 cells) for four amplitudes (gray) plotted with the modeling results for average predicted electric field (blue) and fields delivered with 50% efficiency (red). Shaded regions for the experimental results indicate the standard error of the means. To see this figure in color, go online.

Using the predicted electric field obtained from the FDTD simulation and the membrane-charging model, the resultant fluorescence model overestimates the directly observed membrane charging obtained from the streak kymographs. The model assumes that all the energy, as predicted by the FDTD model using applied voltage measured across a parallel resistor, reaches the cells as an electric field; however, ohmic losses before the electrodes, in the electrodes, or at the electrode-solution interface could decrease the electric field at the cell relative to its nominal value. Additionally, some energy may be lost in the formation of chemical species, such as the formation of hydrogen bubbles through electrolysis, from the electrical discharge from the tungsten electrodes (27). We have observed this effect through the formation of microbubbles during pulsing, albeit at higher potentials. Thus, if we estimate 50% efficiency of average electric-field delivery to the cells (40% of maximal field amplitude) (Fig. 3, d and e), the predicted and observed charging potentials closely match, both in membrane charging and discharging times (Fig. 4, red traces). Additionally, the experimentally obtained changes in fluorescence at the end of the pulse plotted against the modeled membrane-charging potential for each exposure were found to be a close fit to the calibration for Fluovolt (Fig. 3 f).

Questions do remain regarding the saturation of the Fluovolt signal for the higher applied voltages used in these experiments. First, from the patch-clamp experiments, the fluorescence signal shows a sigmoidal response from −1 to 1 V, with the signal becoming more nonlinear beyond ±200 mV (Fig. S2, f–h). This nonlinearity appears to contribute, at least partially, to the observed saturation of fluorescence signal. Second, we considered that the electric field itself may be saturating in the solution. However, a number of studies, using similar electric pulse exposure systems, have shown a proportional increase in uptake of markers of membrane permeabilization, such as propidium ion and Yo-Pro-1, on applied voltage at amplitudes far beyond those used in this study (24, 28, 29). The electric-field dependence was also corroborated by whole-cell patch-clamp experiments in which the membrane resistance was reduced proportional to the applied energy to the system (30, 31). Additionally, although the maximal fluorescence change observed from the streak camera experiments was a close match to that observed in patch clamp for negative membrane potentials, the percentage of change for positive membrane charging was below (32 vs 62%) that observed in patch clamp (Fig. 2 h). This result may indicate that, for positive potentials, the membrane is breaking down at charging potentials of ∼300 mV and preventing any further membrane charging in these areas. The idea of breakdown potential has been extensively discussed for studies involving electroporation and nanosecond electric pulses, although the predicted membrane breakdown voltage is typically closer to 1 V (24, 32, 33).

Conclusions

The use of a streak camera imaging in conjunction with Fluovolt resolves rapid changes in membrane potential below membrane-breakdown voltages. Fluovolt responds to changes in membrane potential with temporal resolutions of ∼25 ns and appears unperturbed by the flux of the laser during the brief imaging period. This, to our knowledge, novel approach can be used to monitor changes in membrane potential from external stressors, such as applied electric fields, opening dynamics of ion channels, and dynamic responses in membranes to various environmental conditions. Our approach offers superior dynamic resolution with the potential to record single rapid events, avoiding sample variability and shot-to-shot repeatability that plagues previous techniques. Our method also retains spatial information by allowing for simultaneous detection (in one spatial dimension) of multipoint membrane potential distributions across a biological sample. Our approach provides electrophysiological-like recordings of membrane potential with higher temporal resolution and spatial information from opposing sides of a single cell. Furthermore, the approach is contact free, not requiring perturbation of the cell with a patch electrode, which allows for rapid throughput measurements of cell-membrane responses to a variety of exposure conditions without dialysis of the cell.

Author Contributions

H.T.B. and B.L.I. conceived and designed the experiments. H.T.B., B.L.I., C.C.R., J.N.B., and A.V.S. performed the experiments. H.T.B., B.L.I., and A.V.S. analyzed the data. H.T.B. implemented and ran the models. H.T.B., B.L.I., and C.C.R. wrote the article.

Editor: Brian Salzberg.