Action spectra and mechanisms of (in)efficiency of bipolar electric pulses at electroporation

Abstract

The reversal of the electric field direction inhibits various biological effects of nanosecond electric pulses (nsEP). This feature, known as “bipolar cancellation,” enables interference targeting of nsEP bioeffects remotely from stimulating electrodes, for prospective applications such as precise cancer ablation and non-invasive deep brain stimulation. This study was undertaken to achieve the maximum cancellation of electroporation, by quantifying the impact of the pulse shape, duration, number, and repetition rate across a broad range of electric field strengths. Monolayers of endothelial cells (BPAE) were electroporated in a non-uniform electric field. Cell membrane permeabilization was quantified by YO-PRO-1 (YP) dye uptake and correlated to local electric field strength. For most conditions tested, adding an opposite polarity phase reduced YP uptake by 50–80%. The strongest cancellation, which reduced YP uptake by 95–97%, was accomplished by adding a 50% second phase to 600-ns pulses delivered at a high repetition rate of 833 kHz. Strobe photography of nanosecond kinetics of membrane potential in single CHO cells revealed the temporal summation of polarization by individual unipolar nsEP applied at sub-MHz rate, leading to enhanced electroporation. In contrast, there was no summation for bipolar pulses, and increasing their repetition rate suppressed electroporation. These new findings are discussed in the context of bipolar cancellation mechanisms and remote focusing applications.

1. Introduction.

Nanosecond electric pulses (nsEP) are a new modality for clinical applications such as cancer and tissue ablation, immune and neuroendocrine stimulation, and defibrillation[1–6]. Primary effects of nsEP are direct stimulation and nanoelectroporation of the cell plasma membrane and of intracellular organelles[7–10]. Nanopores are distinguished by complex ion channel-like properties, including voltage sensitivity, cation selectivity, and inward rectification[11–15]. They are remarkably stable and remain open for minutes, affecting the resting transmembrane potential (TMP) and ion gradients. nsEP can modify membrane proteins, causing lasting changes in the activity of endogenous ion channels[16, 17]. Stimulation by nsEP can bypass membrane receptors and ion channels to elicit second messenger Ca²⁺ and PIP₂ signaling and evoke neuromediator release and other downstream effects[7, 12, 18–23]. Intense nsEP treatments cause cytoskeleton rearrangements[19, 24], osmotic stress, cell swelling and blebbing[12, 19, 23, 25], and apoptotic or necrotic cell death[23, 26–28].

The electric field reversal weakens diverse nsEP effects[29, 30]. As a result, a bipolar nsEP may be less effective than a single phase of the same pulse. This phenomenon, known as “bipolar cancellation,” was reproducibly observed by independent groups[31–35] and has recently been employed for interference focusing of nsEP effects at a distance from electrodes[36]. The challenge of remote focusing, which is to avoid the effects in the strongest electric field near electrodes, was addressed by using two pairs of electrodes to deliver two synched bipolar nsEP with a phase shift. Bipolar nsEP are inefficient near the electrodes, despite the high electric field strength. However, they superimpose remotely into a unipolar pulse, which is biologically-efficient despite the inevitable electric field weakening with distance. In other words, the transition from bipolar to unipolar pulse shape cancels the bipolar cancellation (“CANCAN effect”[36]).

Prospective applications of CANCAN range from non-invasive deep brain stimulation to precise tumor ablation with minimal healthy tissue damage. These applications will rely on the proper choice of nsEP parameters to prevent focal effects near electrodes. However, studies which explored the dependence of cancellation on the pulse shape, duration, and inter-phase delay produced surprisingly contradictory results. For instance, electroporation of CHO cells and cardiomyocytes by “long” nsEP (100s of nanoseconds) was canceled best when the 2^nd phase of the pulse was smaller than the first one in either amplitude[37] or duration[32]. Also, longer bipolar nsEP (830+830ns) were more efficient at canceling propidium uptake into electroporated cardiomyocytes than shorter 200+200ns pulses[37]. In contrast, nerve excitation was canceled most efficiently when the opposite nsEP phases were identical[38] and shortening of nsEP enhanced cancellation[37, 39].

Instead of cancellation, symmetrical bipolar nsEP (150 to 400 ns per phase) enhanced Ca²⁺ influx into adrenal chromaffin cells[31]. In these cells, cancellation could only be achieved with asymmetrical bipolar pulses[31] or with very short (2 ns per phase) symmetrical pulses [34]. When the 2-ns phases were split apart and delivered as two opposite polarity pulses, cancellation gradually weakened and vanished as the inter-pulse interval reached 30 ns. A similar inter-pulse limit of 30 ns was reported for cancellation of electroporation in U87-MG glioblastoma cells with 10-ns pulses[40]. However, electroporation of CHO cells by longer 60- and 300-ns pulses could still be inhibited with inter-pulse intervals as long as 10–50 μs[29, 37, 41]. Nerve stimulation by 700-ns pulses was partially canceled at up to 200-μs pulse separation [39]. Moreover, a single study reported cancellation of electroporation when a 300-ns pulse was delivered 10 ms and even 1 s after the preceding 900-ns pulse of the opposite polarity[41].

These profound differences suggest that bipolar cancellation is not a single phenomenon but rather an “umbrella” term for diverse effects with qualitatively different underlying mechanisms. For example, cancellation of nerve excitation was fully consistent with the accelerated discharge mechanism[29, 38, 39], while cancellation of electroporation was not[35, 42]. The data also indicate that varying one pulse parameter at a time may not be sufficient to test the mechanisms and identify the best conditions for bipolar cancellation. Therefore, here we explored the impact of the pulse shape, duration, number, and repetition rate on electroporation and its cancellation across a broad range of electric field strengths. This task was facilitated by applying a non-uniform electric field to cell monolayers, so that a single nsEP treatment yielded cancellation measurements for multiple electric field strengths.

2. Materials and methods

2.1. Cell culture

Bovine pulmonary artery endothelial (BPAE) cells were a gift from Dr. J. Catravas (Center for Bioelectrics, ODU). Cells were grown as described recently [36], at 37°C with 5% CO2 in air in a low-glucose DMEM medium with 2.5 μg/ml amphotericin B (Thermo Fisher Scientific, Waltham, MA), 100 I.U./ml penicillin, 0.1 mg/ml streptomycin (Gibco, Gaithersburg, MD), and 10% fetal bovine serum (Atlanta Biologicals, Norcross, GA). Cells were harvested 12–18 h before experiments and transferred at 2 ml/dish, (0.3–0.6) × 10⁶ cells/ml, into 35 mm glass bottom culture dishes (MatTek, Ashland, MA) pre-coated with 0.2% gelatin. Dishes with 80–90% confluent monolayers were taken out of the incubator 20–25 min before nsEP exposure, and the growth medium was replaced with 2 ml of a physiological solution (PS) containing (in mM): 140 NaCl, 5.4 KCl, 2 CaCl₂, 1.5 MgCl₂, 10 D-glucose, and 10 HEPES (pH 7.4, 290–300 mOsm/kg, 1.6 S/m). The chemicals were from Sigma-Aldrich (St. Louis, MO).

For strobe imaging experiments only (Sections 2.6 and 3.6), we used Chinese hamster ovary cells (CHO-K1). This method collects data from multiple identical electric pulse exposures to reconstruct membrane charging kinetics from images taken at different delays after pulse. If the membrane state changes substantially from one electric pulse to another (e.g., by activation of voltage-gated channels), this could bias the observations and make their interpretation more complicated. CHO cells, in contrast to BPAE, typically do not express any endogenous voltage-gated channels[43], making them an ideal model that responds uniformly to multiple non-electroporating stimuli[44]. CHO cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA) and propagated in Ham’s F12 K medium (Mediatech Cellgro, Herdon, VA) with same supplements as above except amphotericin B. The day before experiments, cells were seeded at (0.3–0.5) × 10⁵ on 35 mm glass bottom culture dishes.

2.2. Electric pulse stimulation

Pulses with 300-to 1,200-ns duration were delivered to cells by an EPULSUS-FPM4–7 generator built by EnergyPulse Systems (Lisbon, Portugal)[38, 45, 46]. To generate bipolar electric field, positive polarity electric pulses were applied in alternation to two electrodes, and the inactive electrode served for current return (ground). Pulse duration, amplitude, and repetition rate were programmed in the EPULSUS control software. Pulses of longer duration (50 to 500 μs) of both polarities were created using a MOSFET-based generator custom built in-house [11, 47, 48]. The duration of positive and negative phases, pulse repetition rate, and pulse number were controlled with a model 577 digital delay generator (Berkley Nucleonics, San Rafael, CA).

Pulses were delivered to the cell monolayer with a pair of blunt, hollow, stainless-steel needles 1.5-mm in diameter (Integrated Dispensing Solutions, Agoura Hills, CA). They were positioned orthogonal to the monolayer with 3-mm side-to-side separation and lowered using an MX130L micromanipulator (Siskiyou Corporation, Grants Pass, OR) until touching the glass bottom of the culture dish (Fig. 1A).

Fig. 1.

Electroporation of cell monolayers with electric pulses. A: A schematic of a 35-mm glass bottom Petry dish with two electrodes installed orthogonal to a cell monolayer on the bottom. B: Electric field (V/m) in the plane 1 μm above the bottom of the dish produced by 1 V applied to the electrodes. Two black spots are the footprints of the electrodes. C: Electric field distribution along the lines aa’ and bb’ in panel B (a solid blue line and a dashed green line, respectively). D: Sample traces of nano- and microsecond range pulses. E: Heating of thermochromic liquid crystal sheets by trains of 5 bipolar pulses. Pulse repetition rate, phase duration, and the 1^st phase amplitude are labeled in the legends. The second phase had the same duration but 50% smaller amplitude. White dashed line circles mark the location of electrodes. The scale at the right shows the color transition temperatures. F: Sample images of a cell monolayer. Electroporated cells are visualized by YO-PRO-1 staining (green; all panels) and all cells are labeled with MemBrite dye (blue; shown in the right panels only). Dark circles with no electroporated cells are the footprints of electrodes. Bottom panels show a high-resolution image of the area identified by a small rectangle in the top panels. See text for more details.

Pulse shapes and amplitudes were continuously monitored with a TDS 3052 oscilloscope (Tektronix, Beaverton, OR, USA) using a DP20003 High Voltage Differential Probe (Shenzhen Micsig Instruments Co., Guangdong, China) for EPULSUS-FPM4–7 generator and a P2301B high-voltage probe (Qingdao Hantek Electronic, Qingdao, China) for the other pulser. The probes were directly connected to the electrodes. Sample pulse traces are shown in Fig. 1D.

2.3. Numerical simulation of the electric field

The electric field was calculated with a low-frequency finite element solver Sim4Life V5.2 (Zurich Med Tech, Zurich, Switzerland), similarly to what was reported previously[38, 46]. The electrodes were simulated as two parallel stainless-steel cylinders, 1.5 mm in diameter and 3 mm apart. The electrodes were positioned orthogonal to and in touch with the glass bottom of a 35-mm diameter plastic dish. The dish was filled with a 1.6 S/m solution to the depth of 3.5 mm. The model was meshed to approximately 18 million cells, with a minimum step of 30 μm. Electric field was calculated in the plane 1 μm above the glass with a Low-Frequency Electro Ohmic Quasi-Static solver for 1 V applied between the electrodes (Fig. 1B, C). Regions of interest (ROI) to measure membrane permeabilization (section 2.5) were chosen at the equal distance from two electrodes (along the bb’ line), except for a single set of measurements where ROI were placed along the line connecting anode and cathode (aa’ line), to compare anodic and cathodic effects.

2.4. Thermometry

Maximum heating from electric pulse treatments was estimated with R25C5B calibrated thermochromic liquid crystal sheets (LCR Hallcrest, Glenview, IL) as described previously [36, 46]. A sheet was inserted between the electrodes and the glass bottom of the dish filled with PS, and color changes in the film were photographed with a digital camera. Electric pulse parameters were selected for the “worst case” heating scenario, i.e., using the maximum pulse amplitude, duration, and repetition rate tested in actual experiments with cells. Measurements were performed at room temperature (21–22 °C).

Color transitions from black to red, green, blue, and then back to black occur at temperatures of 25±1, 26±1, 30±2, and 44±2 °C, respectively. Fig. 1E shows heating after a train of 5 bipolar pulses, with the 2^nd phase set at 50% of the 1^st one. Pulses applied at 1 Hz (the repetition rate used in all but one sets of experiments) caused only modest heating, not exceeding 30 °C. At a high repetition rate of 833 kHz, heating exceeded 30 °C but never reached the next transition point of 44 °C and remained close to 30 °C (green to blue transition) along the bb’ line (Fig. 1B) where fluorescence measurements were taken.

Overall, temperature stayed below potentially damaging values in all our experiments. However, modulation of the electroporation efficiency by modest heating at the highest electric field strengths and pulse repetition rates could take place.

2.5. Electroporation protocol and data analysis

Membrane permeabilization by electric pulses was quantified by a change in fluorescence of YO-PRO-1 (YP) dye (Thermo Fisher Scientific, Waltham, MA). This dye is poorly permeant into intact cells and is only weakly fluorescent in solution. However, YP becomes brightly fluorescent upon entering the cytosol through electroporated membranes and binding to nucleic acids, thus making it a sensitive marker of electroporation [29, 36, 41, 46].

Cell monolayers typically contain a small percentage of “spontaneously” dead cells whose membrane poses no barrier for YP staining. Such cells fluoresce much brighter than electroporated ones and could distort measurements of YP emission averaged within a ROI. To prevent this artifact, monolayers were pre-treated with propidium iodide (PI; Thermo Fisher Scientific), which bound to the nucleic acids in dead cells. PI has an excitation peak close to the YP emission peak and effectively quenches YP fluorescence when co-localized [23, 46]. Pre-treatment with PI did not fully prevent the subsequent labeling of dead cells with YP, but reduced YP brightness to a level negligible for electroporation measurements.

Another potential inaccuracy in electroporation measurements by YP uptake could come from fluctuations of cell density in the monolayer. To minimize the impact of such fluctuations, cells were stained with MemBrite Fix 405/430 (Biotium, Inc., Fremont, CA), which labels all cells by binding to cell surface proteins. This dye is reportedly non-toxic, does not associate with membrane lipids, and did not appear to change the electroporation efficiency. Normalization of YP fluorescence (electroporated cells) to MemBrite fluorescence (all cells within ROI) minimized the impact of local cell density on quantitative measurements of electroporation.

A culture dish with adhered cells was taken out of the incubator and rinsed twice with PS. Next, cells were incubated for 10 min in PS with 5 μg/ml of PI, rinsed, and labeled with MemBrite following manufacturer’s instructions. Cells were rinsed twice with a fresh PS and left in PS containing 1 μM of YP dye. In 1–2 min, cell monolayers were treated with electric pulses followed by a 10-min incubation in the dark, at room temperature. YP staining was halted by rinsing with fresh PS, and monolayers were imaged with an IX83 microscope (Olympus America, Hamden, CT).

The microscope was configured with an automated MS-2000 scanning stage (ASI, Eugene, OR), X-Cite 110LED illuminator (Excelitas Technologies Corporation, Waltham, MA), and an ORCA-Flash4 sCMOS camera (Hamamatsu, Shizuoka, Japan). Image acquisition, including filter cube selection, stage re-positioning and synchronization with illumination and camera operation was accomplished with CellSens software (Olympus America). A high-resolution image of the entire sample was stitched from 216 regional images taken with a 10x, 0.38NA objective with DAPI and FITC filter cubes for MemBrite and YP emission, respectively (Fig. 1F).

The average fluorescence intensity for each dye was measured with MetaMorph 7.8.13 (Molecular Devices, Foster City, CA) in 200-μm square ROI placed along either aa’ or bb’ lines (Fig. 1B). In the first case (aa’), the electric field was the weakest in the center and gradually increased towards the electrodes; in the latter case (bb’), the strongest electric field in the center gradually decreased towards the edges (Fig. 1C). The gradient of the electric field strength within one 200-μm ROI was <20% along the aa’ and <10% along the bb’ line. The mean electric field strength for each ROI was used for plotting the respective fluorescence measurements.

The monolayer area within the footprints of the hollow electrodes was not exposed to the electric field and was used to measure the “spontaneous” YP uptake by cells. The electric field-induced YP emission (YP), the value used throughout this paper, was calculated for each ROI as:

$$YP=Y{P}_{ROI}\times M{B}_0/M{B}_{ROI}-Y{P}_0,$$

where YP_ROI and MB_ROI are the YP and MemBrite fluorescence values averaged over the selected ROI; and YP₀, and MB₀ are the respective values measured within the electrodes’ footprints.

In each set of experiments, exposures using different pulse parameters were applied randomly, and each exposure protocol was tested 3–8 times. Data in graphs are presented as mean values ± standard error of the mean. Grapher 16 (Golden Software, Golden, CO) was used for graph preparation and data fitting. For some datapoints in the graphs, the error bars are smaller than the central symbol and may not be visible. Two-sided Student’s t-test was employed to analyze the significance of differences; p<0.05 was considered statistically significant. Due to multiple datapoints and statistical comparisons, we avoided using special symbols to preserve the clarity of graphs. Instead, the statistical significance can be estimated from the gap between the error bars of the compared groups: A gap exceeding the length of the error bars indicates a significant difference at p<0.05 or better[49].

2.6. Strobe imaging of cell membrane charging by electric pulses

The primary effect of nsEP is the rapid charging of the plasma membrane. Therefore, we attempted to link the electroporation efficiency of nsEP to changes of the transmembrane potential (TMP). We employed strobe imaging synchronized with nsEP exposure to study the nanosecond kinetics of TMP charging and relaxation[44]. This method utilizes pulsed laser fluorescence microscopy where a single short pulse laser flash is delivered at a precise time interval before, during, or after nsEP. Cells are loaded with a voltage-sensitive FluoVolt dye, which responds to TMP changes within nanoseconds[50, 51] and has a high sensitivity of about 10% ΔF/F per 100 mV[51]. The camera shutter opens in advance of and closes after the laser flash, capturing one TMP image per nsEP exposure. Multiple nsEP and laser flash combinations are delivered with programmed time interval increments or decrements, to capture the time course of TMP during and after nsEP.

A block diagram and major features of the strobe imaging setup were described recently [44]. This setup was modified to use an Olympus IX71 microscope equipped with an iXon Ultra 897 back-illuminated EM CCD Camera (Andor Technology, Ireland) and synchronized to the EPULSUS generator. A customized q-switched neodymium-doped yttrium aluminum garnet (Nd:YAG) laser (Quantel USA, Bozeman, MT; http://www.andersonlasers.com/new-arrivals.html) was fired in synchrony with nsEP using a custom LabVIEW software and a Stanford DG645 delay generator (Stanford Research Systems, Sunnyvale, CA). Both the 1064 nm (a primary wavelength for a Nd:YAG laser) and the 532 nm (the 2^nd harmonic) were emitted from the laser (~6-ns flash at up to 250 mJ) onto a non-linear crystal to generate the 355 nm light flash via sum frequency generation. This light was directed via three beamsplitters (part CVI BSR-35–1025, CVI, Albuquerque, NM) and focused into a cuvette containing Coumarin 440 laser dye (Luxottica Exciton, USA) dissolved in methanol, to generate a ~6-ns flash centered around 440 nm. The beamsplitters served to remove any residual 1064 nm light from the optical system. The 440 nm light was captured by a 600 μm core fiber optic (Thorlabs, Newton, NJ) and coupled into the microscope via the rear port to excite FluoVolt dye. The excitation light was additionally cleaned up with a 533 nm notch filter (part NF533–17, Thorlabs Inc., Newton, NJ), which replaced the excitation filter in a standard FITC cube and removed any residual 532 nm laser light collected by the optical fiber. Individual cells or small groups of cells were imaged with a 60x, 1.42 NA objective. In this study, the time interval between nsEP and the laser flash was changed in 25-ns steps to cover a 12.5-μs time interval, which required 500 exposures to nsEP performed at a rate of 6 exposures per second.

Cells were loaded with FluoVolt for 30 min in the dark, in PS containing FluoVolt stock (1/000) and PowerLoad Concentrate (1/100) (Thermo Fisher Scientific). The loading solution was replaced with fresh PS, and the dish with cells was fixed on the microscope stage. Electrodes (a pair of tungsten rods, 0.5-mm diameter, 1-mm side-to-side distance) were lowered into the dish with an MPC-200 manipulator (Sutter, Novato, CA). The electrodes were placed vertically, precisely 50 μm above the bottom of the dish, with the objective’s field of vision centered between them. Electric field at the cells’ location was calculated the same way as described in Section 2.3. To avoid possible membrane damage by repetitive exposures to nsEP, we used a low nsEP voltage of 37 V to generate 0.3 kV/cm electric field at the cell location.

The intensity of the laser dye flash varied within about +/−10%. To account for this variation, FluoVolt signal from the cathode- and anode-facing poles of the cell was scaled to the whole-cell FluoVolt signal (de- and hyperpolarizations at the opposite poles counterbalance each other, and the integrated whole-cell signal is not affected by the external electric field). An additional benefit of scaling to the whole-cell emission was the compensation of FluoVolt bleaching. To achieve even better compensation of bleaching, TMP kinetics in each cell was recorded in two trials. In the 1^st trial, the delay between nsEP and the laser flash was gradually increased, so bleaching increased with the delay. In the 2^nd trial, we started with the maximum delay and gradually decreased it, so bleaching increased for shorter delays. In other cells, we reversed the sequence of the trials. Averaged data of the two trials were considered a single experiment. The mean value for the 40 datapoints before the pulse was taken as 100%. Normalized emission data were averaged across the cell population and plotted as a percent difference from the resting value.

3. Results

3.1. Effect of the 2^nd phase amplitude on electroporation by nsEP

These experiments measured how the electroporation efficiency of bipolar 600+600 ns pulses depends on the ratio of the opposite polarity phases. The voltage of the first phase was kept constant at 5.75 kV, and the voltage of the second phase was set to 0 (making it a unipolar pulse), 25%, 50%, 75%, or 100% of the first phase. A train of 5 pulses applied at 1 Hz evoked YP uptake proportionally to the local electric field strength (Fig. 2A), along with a strong dependence on the 2^nd phase amplitude. While the graphs in Fig. 2A look similar, note the difference in the vertical scale, which is varied from 1200 a.u. for unipolar pulses to only 250 a.u. for 75% bipolar pulses. YP uptake increased exponentially with the electric field, and the presence of the 2nd phase did not change the apparent slope of exponential fits when plotted on a semi-log scale (Fig. 2B). The maximum inhibition of YP uptake, down to ~20% of the unipolar pulses, was achieved with the 2^nd phase set to 50–75% of the first one (p<0.01, Fig. 2C). The degree of inhibition showed no dependence on the electric field strength between 8.7 and 15.1 kV/cm. However, inhibition was weaker at 6.7 kV/cm (p<0.01 compared to the higher electric field strengths for 50% bipolar pulses, and p<0.05 for 100% bipolar pulses). The mechanism why bipolar cancellation was weaker for electric pulses which are just marginally strong to cause electroporation is unclear and remains to be investigated.

Fig. 2.

The addition of the opposite polarity phase to a 600-ns unipolar pulse inhibits electroporation. A: The impact of the second phase amplitude (% to the first phase, see legends) on the electroporation efficiency (measured as YP emission). Cells were electroporated by a train of 5 unipolar 600 ns pulses (left panel) or bipolar 600+600 ns pulses at 1 Hz. Electric filed strength (x-axis) is for the first phase of the pulse. Note the different vertical scale for different panels. Mean ± s.e., n = 3–8. B: The same data combined in a single plot with a semi-log vertical scale. Low YP emission signal data (at less than 6–8 kV/cm) were omitted. Dashed lines are best fits of the data using exponential function. C: The electroporation efficiency plotted against the second phase amplitude (% to the first phase), for several selected field strengths (legends, kV/cm). YP emission after a unipolar pulse exposure (0% second phase) was taken as 100%.

These findings confirm our previous observations[37] that the 2^nd phase of 50–75% is most efficient for bipolar cancellation, and extend them to a different type of cells and to a broad range of electric field strengths. In the subsequent series of experiments, the second phase of bipolar pulses was kept at 50% of the 1^st one to maximize the expected cancellation effect.

3.2. Effect of pulse number

Cells were electroporated by trains of 5, 10, or 15 uni- or bipolar pulses (600 ns and 600+600 ns) at 1 Hz (Fig. 3, A–C). Electroporation by both types of pulses expectedly increased at the higher electric field strengths and for larger pulse numbers. Increasing the pulse number increased YP uptake almost linearly [15]. Therefore, YP uptake plots for 5, 10, and 15 pulses appeared almost identical when plotted with the y-axis scaled proportionally (to 800, 1600, and 2400 a.u., respectively).

Fig. 3.

Effect of the number of pulses on electroporation by unipolar and 50% bipolar pulses (filled and open symbols, 600 ns and 600+600 ns, respectively). A, B, and C: Trains of 5, 10, or 15 pulses, respectively, all at 1 Hz. Note that the y-axis is scaled proportionally to the pulse number (800, 1600, and 2400 a.u.). Mean ± s.e., n = 3. Solid lines are the best fits using the 2^nd degree polynomial function. D. The efficiency of bipolar pulses relative to that of unipolar pulses. Solid lines are the ratios of the best fits for bi- and unipolar pulses in panels A-C. See Fig. 2 and text for more detail.

For all nsEP treatments, the increase of YP emission with the electric field strength was fitted with the 2^nd degree polynomial function. The ratio of the best fits for bi- and unipolar pulses shows how strongly the addition of the 2^nd phase inhibits electroporation (Fig. 3D). The data where YP uptake could not be reliably separated from the background (<7–8 kV/cm) were not included. The plots for 5, 10, and 15 pulses nearly overlapped, showing that within the studied limits cancellation of electroporation by adding a 50% opposite polarity phase did not depend on the pulse number or the electric field strength. This result is consistent with some earlier data [37] but partially contradicts the reported dependence of cancellation on the electric field strength [36]. The difference could possibly be explained by using triphasic bipolar pulses (which caused stronger cancellation, up to 5-fold) and 25 pulses/train (so that YP uptake could be reliably measured already at 4–6 kV/cm) in the previous study [36]. In the subsequent experiments, we utilized a constant number of 5 pulses per train.

3.3. Effect of pulse duration

As discussed in the Introduction, the dependence of bipolar cancellation on pulse duration is one of the most controversial topics. Here we compared electroporation by 1-Hz trains of 5 uni- or bipolar pulses, with phase durations of 300 ns, 600 ns, 1200 ns, 50 μs, 100 μs, or 500 μs (for unipolar pulses, this was the entire pulse duration). To generate a measurable electroporation area, the voltage of the first phase was set to 5.75 kV for all three nsEP durations and to 425V, 350V, and 250V, respectively, for the microsecond range pulses. The voltage of the second phase was fixed at 50% of the first one.

The addition of the second phase profoundly inhibited electroporation by nsEP of all three durations (Fig. 4,A–C). Electroporation by 300+300 and 600+600 ns pulses was at 20–25% of the respective unipolar pulses and did not depend on the electric field strength. However, longer 1200+1200 ns pulses were somewhat more efficient at cancellation at moderate electric field strengths, reducing the YP fluorescence 10-fold (Fig. 4D).

Fig. 4.

Effect of the pulse duration on electroporation by unipolar and 50% bipolar pulses (filled and open symbols, respectively). All exposures were to 5 pulses at 1 Hz. Legends mark the duration of unipolar pulses (or one phase duration for bipolar pulses). To produce a measurable electroporation area with long pulses (E-G), the voltage of the applied pulses was reduced compared to nsEP (A-C), and the range of the electric field strength (x-axis) changed accordingly. Solid lines in panels A-C and E-G are the best fits using either exponential or polynomial function. Panels D and H show the efficiency of bipolar pulses relative to that of unipolar pulses. Solid lines are the ratios of the best fits for bi- and unipolar pulses in panels A-C and E-G. Mean ± s.e., n = 3–8. See Fig. 2 and text for more detail.

Surprisingly, the addition of the second phase reduced the efficiency of “long” microsecond pulses as well, even at 100 and 500 μs phase durations (Fig 4. E–G). Cancellation by these longer pulses was relatively weak (to 30–50% of the unipolar pulse effect) but highly significant (p<0.01). The trend for more efficient cancellation at the lower electric field strengths persisted throughout the microsecond range (Fig. 4H).

The time intervals of 50–500 μs are too long to explain bipolar cancellation by the assisted membrane discharge or a similar process. The existence of bipolar cancellation at these intervals suggested a long-lived mechanism able to support YP uptake after the first phase of the pulse has ended. We hypothesized that this mechanism could be related to the formation of toxic chemical species at one of the electrodes. The formation of toxic electrochemical byproducts is negligible with nanosecond-range pulses but becomes a significant factor for stimulation and electroporation with conventional “long” pulses [52–54]. A recent study found that ions released from a stainless-steel anode could make the culture medium cytotoxic[55]. Bipolar waveforms are commonly used to reduce electrolytic contamination and its bioeffects[56]. Thus, bipolar cancellation for 50–500 μs pulses could be due to the reduction of electrolytic contamination and its toxic effects.

To test this idea, we re-analyzed the data by placing ROI along the line connecting anode and cathode (line aa’ in Fig. 1B, C). Membrane charging does not differ for cells near anode and near cathode, as long the electric field strength at these locations is the same. Thus, YP uptake along the aa’ line should be symmetrical for electroporation itself, and any asymmetry will indicate a different (electrochemical) reaction affecting cells at one of the electrodes. YP uptake data measured along the aa’ line was nearly symmetrical for 300- and 600-ns pulses (Fig. 5A, B). At 1200 ns, the effect of unipolar pulses appeared marginally stronger at the anodic side (Fig. 5C), and still longer 50–500 μs pulses showed significantly stronger YP uptake (p<0.01) at the anodic site. These results support the idea that bipolar cancellation in the case of 50–500 μs was likely due to the reduction of the toxic impact of chemical species formed at the anode, and are consistent with the study that specifically pointed to the formation of toxic species at the stainless-steel anode[55]. In contrast to the “long” pulses, bipolar cancellation with nsEP resulted from some unidentified mechanisms other than electrochemical reactions at the anode.

Fig. 5.

Asymmetrical enhancement of YP dye uptake near anode after electroporation with “long” microsecond pulses (D-F) but not with shorter nanosecond pulses (A-C). The data from experiment used for Fig. 4 were re-analyzed, by placing regions of interest for YP emission measurements along the line connecting the electrodes (aa’ line in Fig. 1B). X-axis shows both the distance (mm) from the geometrical center between the electrodes and the electric field strength (kV/cm) at the respective distances. Other designations and details are the same as in Fig. 4.

3.4. Effect of pulse repetition rate

Increasing the repetition rate from 1 Hz (which was used in all experiments described above) into kHz and sub-MHz range had complex and opposite effects on electroporation by uni- and bipolar pulse trains (5 pulses, 600 ns or 600+600 ns; the 2^nd phase amplitude set to 50% of the 1^st one).

Unipolar pulses at 10 to 500 kHz showed the same efficiency as at 1 Hz (Fig. 6A,D). However, increasing frequency to 833 kHz sharply enhanced electroporation due to the so-called “MHz compression” effect. YP uptake after 833 kHz exposure was 2–3 fold higher than at other frequencies throughout the studied range of the electric field strengths. The increased efficiency is usually explained by a stepwise build-up of the induced membrane potential when the inter-pulse interval is too short for the relaxation of the induced potential[57, 58].

Fig. 6.

Effect of the pulse repetition rate on electroporation by trains of five uni- or bipolar pulses (600 ns and 600+600 ns, respectively). Panels A-C and D-F show the same YP uptake data plotted against the pulse repetition rate or against the electric field strength, respectively. For bipolar pulses, the electric field strength is that of the first phase, and the 2^nd phase is set 50% smaller. In panels A and D, the data for unipolar pulse trains are labeled with the repetition rate (kHz and Hz) and the electric field strength (kV/cm), respectively. YP uptake following the bipolar pulse exposure was much weaker, so the curves are expanded and labeled in panels B and E on a different vertical scale. Solid lines in panels A,B, D, and E are the best fits using either exponential or polynomial function. Panels C and F show the efficiency of bipolar pulses relative to that of unipolar pulses. Solid lines in C and F are the ratios of the best fits for bi- and unipolar pulses in panels A and B, and D and E, respectively. Mean ± s.e., n = 3 – 8.

In contrast, the efficiency of bipolar pulses at all kHz frequencies was less than at 1 Hz, and the electroporative uptake of YP was just marginally above the background level (Fig. 6B,E). The difference between 1 Hz and kHz-rate trains was maximum (2.5 times) at the highest electric field strength. The reason for the reduced efficiency of kHz bursts is not clear. Perhaps the 5-s total treatment duration at 1 Hz was long enough for electrosensitization (thereby enhancing the effect compared to brief kHz-rate treatments), although electrosensitization was previously reported only for treatments lasting 10 s and longer [59].

To summarize, increasing pulse repetition rate enhanced electroporation by unipolar pulses, but inhibited electroporation by bipolar pulses. This resulted in the strongest bipolar cancellation we have seen thus far (Fig. 6, C.F). At 12–15 kV/cm, the effect of bipolar pulses at 833 kHz was only 3–4% of the effect of unipolar pulses (~30-fold reduction). Bipolar cancellation that strong may translate into major refinements of remote focusing of stimulation and electroporation by nsEP[36].

3.5. Membrane charging by uni- and bipolar pulses at sub-MHz repetition rates

The increase of electroporation efficiency at 833 kHz for unipolar but not bipolar pulses prompted us to explore the charging kinetics of cell membrane by these two modalities. We employed strobe imaging with FluoVolt dye [44] to reconstruct the membrane potential build-up and relaxation with nanosecond resolution. Kinetic data were collected from 18 cells exposed to unipolar 600-ns pulses at 0.3 kV/cm and 17 cells exposed to bipolar 600+600 ns pulses at 0.3 and 0.15 kV/cm for the first and the second phases, respectively (Fig. 7). Unipolar nsEP caused fast membrane depolarization at the cathode-facing pole of the cell, and a mirroring hyperpolarization at the anode-facing pole. Bipolar nsEP caused the same initial polarization during the first phase of the pulse, followed by a faster discharge and an overshoot into the opposite polarity.

Fig. 7.

Time kinetics of the membrane potential at the cathode- and anode-facing poles of a cell (top and bottom plots, respectively) induced by a train of 5 pulses at 833 kHz. The membrane potential is expressed as a percent change in FluoVolt emission (see text for details). The trains were composed of either 600-ns unipolar pulses (traces shown in the lower plot) or 600+600 ns bipolar pulses. The respective TMP changes are identified by the legends. The electric field strength was 0.3 kV/cm for unipolar pulses and for the first phase of bipolar pulses (the 2^nd phase was 50% smaller). Solid lines are the TMP values averaged from 18 or 17 cells (for uni- and bipolar nsEP, respectively) at 25-ns time increments; the standard error of the mean is shown by a lighter “shadow”. Dotted lines at the top panel are the best fits of the charging and discharging kinetics for the 1^st pulse in a train of unipolar pulses, using a single exponential function. See text for more details.

Electroporation efficiency was quantified for a variety of uni- and bipolar pulses

Bipolar electric pulses were 2- to 30-fold less efficient than unipolar ones

The peak 30-fold difference was achieved with sub-MHz trains of 600-ns pulses

Electroporation efficiency was correlated to nanosecond kinetics of membrane charging

We used FluoVolt emission changes at the cathodic pole to calculate the time constants of membrane charging and discharging. The best fit for membrane charging with an exponential function (using the first pulse in the unipolar pulse train) yielded the time constant of 200 ns and the maximum membrane polarization at 31% emission change (425 datapoints, correlation coefficient 0.93, p<0.0001). The best fit for the discharging portion of the curve yielded the same initial polarization (31.5% emission change) but, unexpectedly, a 2.5 times larger time constant of 520 ns (425 datapoints, correlation coefficient 0.85, p<0.0001; see dotted lines in Fig. 7). Fitting the membrane potential for the next pulses in the train consistently produced similar differences between the charging and discharging time constants (data not shown). While this difference was intriguing, the analysis of its underlying mechanisms was beyond the scope of this study.

With the charging time constant of only 200 ns, even the first 600-ns pulse (= 3x the time constant) charged the membrane to 95% of the maximum value. Although charging by the subsequent pulses in unipolar pulse trains started with the partially charged membrane, it reached the same peak potential. A stepwise build-up of the peak membrane potential, the mechanism used to explain the high efficiency of nsEP at near-MHz frequencies[38, 57, 58], did not occur in our experiments with 833-kHz, 600-ns unipolar pulses. The effects of individual nsEP “merged” into one continuous sawtooth-shaped membrane polarization, and this temporal summation was apparently the reason why 833-kHz unipolar pulse trains were more efficient than trains at lower frequencies. Reducing the frequency just to 500 kHz would allow for a 93% discharge between the pulses (assuming the 500-ns time constant and 1400-ns interpulse interval) and make the temporal summation unlikely.

One may speculate that continuously maintaining a certain level of membrane polarization was essential for stabilizing membrane pores and preventing their annihilation[60]. Bipolar pulses caused repetitive alternation between positive and negative membrane potentials, which could facilitate pore closure beyond just returning the potential to the base level. The exact mechanism and kinetics of this type of bipolar cancellation remain to be elucidated.

4. Discussion

Our study was the first to compare electroporation efficiency of uni- and bipolar pulses by varying pulse duration and number, phase ratio, and repetition rate across a wide range of electric field strengths. Only one of these parameters, pulse number, changed electroporation efficiency of uni- and bipolar pulses similarly (Fig. 2), i.e., did not affect bipolar cancellation. Еlectric field strength had a complex effect, which depended on other parameters. For high-frequency nsEP bursts, stronger electric field enhanced cancellation (Fig. 6C, F). In contrast, for 50- to 500-μs long electric pulses, and possibly including 1200-ns pulses, cancellation was somewhat stronger with the weaker electric field (Fig. 4D, H). There was no such effect for shorter 300- and 600-ns pulses. We showed that electroporation by microsecond-range pulses was significantly stronger at the anode than at the cathode, indicating the role of chemical byproducts forming at the anode in injuring cell membranes (or inhibiting the membrane repair; Fig. 5). The reduced effect of bipolar pulses was simply a result of reducing the yield of toxic electrochemical byproducts, a mechanism that we usually would not refer to as “bipolar cancellation” (although formally it qualifies as such).

The ratio of pulse phases and repetition rate were two parameters that had a highest impact on the bipolar cancellation efficiency. As in the previous study[37], maximum cancellation was achieved when the 2^nd phase was at 50–75% of the first one, and this dependence was observed throughout a wide range of the electric field strength (Fig. 1B, C). This dependence was observed for adherent (CHO) and suspension-based (U937) cell lines, as well as for primary ventricular cardiomyocytes. The dependence of cancellation on the electric field strength was qualitatively similar in three adherent cell lines (CHO, BPAE, and HEK)[36]. Cancellation was relatively weak or absent at the lowest electric field strengths. However, maximum cancellation was achieved at a different electric field strength for each of the three cell lines. Our ability to extrapolate and quantitatively predict cancellation efficiency in diverse cell types will likely remain limited until the biophysical mechanisms responsible for cancellation are fully established.

Of particular interest is the enhancement of bipolar cancellation at high repetition rates (Fig. 6), thanks to the enhancement of electroporation for unipolar pulses concurrently with its inhibition for bipolar pulses. At 833 kHz, we achieved a nearly 30-fold difference in bi- and unipolar pulse trains. The kinetic analysis of the induced membrane potential (Fig. 7) revealed that the enhanced electroporation at this sub-MHz rate results from the temporal summation of individual nsEP to produce a continuous membrane polarization. However, with the employed nsPEF duration (which happened to be threefold the membrane charging time constant), membrane polarization reached nearly maximum during the first pulse in the train and had no room to grow further in response to the subsequent pulses. Shortening the pulse duration will be a straightforward solution to take advantage of this mechanism for unipolar pulses while the (in)efficiency of shorter bipolar pulses will likely stay unchanged. This way we can plausibly expect more than a 100-fold cancellation, with major implications for focusing nsEP effects remotely from electrodes[36].