Excitation and electroporation by MHz bursts of nanosecond stimuli

Abstract

Intense nanosecond pulsed electric field (nsPEF) is a novel modality for cell activation and nanoelectroporation. Applications of nsPEF in research and therapy are hindered by a high electric field requirement, typically from 1 to over 50 kV/cm to elicit any bioeffects. We show how this requirement can be overcome by engaging temporal summation when pulses are compressed into high-rate bursts (up to several MHz). This approach was tested for excitation of ventricular cardiomyocytes and peripheral nerve fibers; for membrane electroporation of cardiomyocytes, CHO, and HEK cells; and for killing EL-4 cells. MHz compression of nsPEF bursts (100-1000 pulses) enables excitation at only 0.01-0.15 kV/cm and electroporation already at 0.4-0.6 kV/cm. Clear separation of excitation and electroporation thresholds allows for multiple excitation cycles without membrane disruption. The efficiency of nsPEF bursts increases with the duty cycle (by increasing either pulse duration or repetition rate) and with increasing the total time “on” (by increasing either pulse duration or number). For some endpoints, the efficiency of nsPEF bursts matches a single “long” pulse whose amplitude and duration equal the time-average amplitude and duration of the bursts. For other endpoints this rule is not valid, presumably because of nsPEF-specific bioeffects and/or possible modification of targets already during the burst. MHz compression of nsPEF bursts is a universal and efficient way to lower excitation thresholds and facilitate electroporation.

Graphical Abstract

1. Introduction

High intensity nanosecond pulsed electric field (nsPEF) is a new and unique technique for neuromuscular stimulation[1-6], mobilization of cytosolic Ca²⁺ in both excitable and non-excitable cells[1, 7-9], and opening of inwardly rectifying and ion-selective stable nanopores in cell membrane[10, 11]. nsPEF treatments cause non-chemical activation of diverse cell and tissue types and, at higher doses, lead to highly selective cell killing by necrotic and apoptotic pathways[12-14]. nsPEF is unique in its ability to disrupt intracellular membranes[1, 8] and to cause so-called “bipolar cancellation”[15], which was recently utilized to time the opening of voltage-gated sodium channels[6]. Existing and emerging applications of nsPEF range from membrane biophysics to defibrillation, peripheral nerve and deep brain stimulation, and cancer ablation.

A major limitation to these and other applications is the need for high pulse voltages, in order to exceed the electric field (EF) threshold for short pulse durations. This threshold increases with pulse shortening, up to tens of kV/cm. Extension of a strength-duration curve for neurostimulation into nanosecond range yields thresholds of about 1 and 240 kV/cm for 100- and 1-ns pulses, respectively[16]. Some other reported thresholds for a single nsPEF stimulus are 1.4-2.4 kV/cm for activation of cardiomyocytes by 200-ns pulses[17, 18]; 1.8 kV/cm for induction of calcium transients in HEK293 cells with 300-ns pulses[9]; 6 kV/cm and 1 kV/cm for permeabilization of CHO cells by 60- and 600-ns pulses, respectively[19]. Pulse voltage required to reach these thresholds can already be prohibitively high: For example, achieving 10 kV/cm to ablate a tumor between two parallel-plate electrodes with a 2-cm gap requires 20 kV applied to the electrodes. Such voltages are beyond the capability of most nsPEF generators and may present a high-voltage hazard.

Delivering multiple stimuli can result in a stepwise voltage build-up on the membrane of the target cell, eventually reaching the excitation or electroporation threshold when the interpulse interval is shorter than the relaxation of the induced transmembrane potential [20, 21]. Charging time constants in mammalian cells are typically at 0.1-1 μs[1, 22]. Temporal summation can only be expected at interpulse intervals smaller than 3-5 time constants (which correspond to 95 and 99% discharge between sequential pulses), which translates into repetition rates from tens of kHz to more than 1 MHz. Indeed, delivering multiple nsPEF at repetition rates of 1 Hz-5 kHz causes stronger effects than a single pulse, but either without decreasing the threshold[23], or with a modest reduction of the threshold[24] (which could in fact reflect a better detection of stronger effects). Much higher repetition rates needed for temporal summation are beyond the capabilities of most nsPEF generators, hence the potential of this approach to lower stimulation and electroporation thresholds has been scarcely explored. A handful of recent studies which systematically looked at electroporation at nsPEF repetition rates up to 500 kHz[25] and 1 MHz[20] reported complex changes which also varied with the electroporation marker employed. Nonetheless, facilitation of electroporation [20] and gene transfer[21] at near-MHz rates confirmed the concept and the achievability of temporal summation of nsPEF stimuli.

Here we show that MHz compression of nsPEF bursts is a powerful and universal approach to facilitate excitation and electroporation and to lower their thresholds. We were able to evoke nsPEF bioeffects at EF levels of just 10-150 V/cm, which is one-two orders of magnitude lower than reported previously. We also explored how nsPEF efficiency depends on pulse and burst parameters and compared the effects to conventional (“long”) electric stimuli

2. Materials and Methods

2.1. Cells

We used adherent cell lines HEK 293 (human epithelial kidney), CHO-K1 (Chinese hamster ovary) and mouse ventricular cardiomyocytes (VCM) for time-lapse fluorescence imaging assays, and suspension based EL-4 cells (mouse lymphoma) for viability studies. The cell lines were obtained from ATCC (Manassas, VA) and propagated as recommended by the supplier. VCM were isolated from adult DBA/2 J mice by enzymatic digestion during Langendorff perfusion, following protocols reported in detail recently [26, 18]. Animal protocols were approved by the Institutional Animal Care and Use Committee, and experiments were performed in accordance with relevant guidelines and regulations. VCM were seeded on laminin-coated 10-mm glass cover slips and used in experiments within 48 h.

2.2. Recording of nerve compound action potentials (CAPs)

Nerve isolation (n. ischiadicus + n. peroneus) from the bullfrog Rana catesbiana and CAP recording followed procedures reported recently[6]. Animal protocol was approved by ODU Institutional Animal Care and Use Committee. Isolated nerve was ligated at both ends and submerged in a chilled physiological solution containing (mM): 140 NaCl, 5.4 KCl, 1.5 MgCl₂, 2 CaCl₂, 10 glucose, and 10 HEPES (pH 7.3, 290–300 mOsm/kg, 1.6 S/m). CAPs were elicited with different generators described below and recorded with an MP160 Data Acquisition System (BIOPAC Systems, Goleta, CA); see Supplementary Material for the set-up, stimulation, and dosimetry details.

2.3. Electric pulse generators

For most experiments, we used bursts of 5 to 1000 pulses at repetition rates from 1 Hz to 3-4 MHz, with individual nsPEF duration ranging from 11 to 500 ns. We also used single pulses of up to 1 ms in duration to compare bioeffects with nsPEF bursts. To deliver such diverse stimuli into different biological loads (impedance from 8 to 200 ohm), we employed several high-power nsPEF generators custom-built in-house at ODU and at Vilnius Gediminas Technical University in Vilnius, Lithuania [27], and a low-power model 577 digital delay generator (Berkley Nucleonics, San Rafael, CA). The delay generator was most flexible for setting pulse parameters, but the output pulse amplitude was limited to only 20 V into a 200-ohm load. Each of the custom-built generators had limitations in the burst duration and minimum pulse duration, but could deliver up to 3 kV in 100-200 ohm loads (such as adherent cells on a coverslip), or up to 500 V into 8-10 ohm loads (such as an electroporation cuvette with cell growth medium). Except for cuvette exposure, all pulses were unipolar and nearly rectangular, with rise and fall times <15% of pulse duration (Fig. 1A, inset). For cuvette exposure, long rise and fall times at the employed setting of 200 ns duration (at 50% height) resulted in a triangular pulse shape. For generation of single, nearly rectangular, unipolar “long” pulses (hundreds of microseconds) we utilized either a model S88 stimulator (Grass Instrument Co., Quincy, MA) or a custom-built MOSFET-based generator [28]. Pulse shapes and amplitudes were controlled with a TDS3052 oscilloscope (Tektronix, Beaverton, OR).

Fig. 1. Nerve excitation with nsPEF bursts.

(A) The effect of pulse repetition rate on the threshold of excitation by bursts of 340-ns pulses, 5 or 100 pulses/burst. The inset is a sample pulse shape at 60% duty cycle. (B) Threshold time-average EF decreases with burst duration independently of indicated nsPEF duration (“11-18 ns” is the duration range for the shortest setting). (C) Comparison of excitation thresholds for single pulses of indicated duration and nsPEF bursts; see text for details. Mean ± s.e., n=5-9.

2.4. Cell stimulation and permeabilization.

Cell response to nsPEF was monitored by time-lapse fluorescence imaging as described previously, to detect either changes of the membrane potential with a FluoVolt dye[4, 18], or increases in cytosolic Ca²⁺ with either Fluo-4 or Fluo-8 dye [29, 17], or YO-PRO-1 dye uptake[23]. The membrane potential and Ca²⁺ indicators were pre-loaded into cells [4, 29, 17, 18], whereas YO-PRO-1 was added to the bath solution at 1 μM throughout the experiment. The bath solution was either the physiological solution defined above, or (when indicated) the same solution mixed 1:9 with an isosmotic sucrose solution, to decrease conductance and facilitate electroporation. A pair of tungsten rod electrodes (100-μm diameter, 140-170 μm gap) connected to a pulse generator were positioned within the microscope field of vision so that the selected cell (or a small group of cells) was centered between their tips [29, 17, 18]; then the electrodes were lifted precisely to 50 μm above the coverslip surface. The pulsed power system was triggered and synchronized with image acquisition by a TLL pulse protocol using a Digidata 1440A board and Clampex software (Molecular Devices, Foster City, CA). EF was calculated by 3D numerical simulations using a finite element solver COMSOL Multiphysics (Stockholm, Sweden)[4, 18].

2.5. Cell viability assay

EL-4 cells where re-suspended in the growth medium (DMEM with 10% FBS) at 1.2 10⁶/ml, and 100-μl aliquots where placed in 1-mm gap electroporation cuvettes. Burst of nsPEF were applied at room temperature; maximum (adiabatic) heating from the exposure was calculated as described elsewhere[19] and did not exceed 6 °C. Cells were returned to the incubator, and viability was measured in 24 h with Presto Blue metabolic assay (ThermoFisher Scientific, Waltham, MA)[30].

3. Results and discussion

3.1. Lowering nsPEF stimulation threshold for a peripheral nerve

Nerve fibers can be repeatedly excited by nsPEF without damage[3], making them a proper model to study temporal summation. Large charging time constant (on the order of 10-400 μs [31-33]; compare to 0.1-1 μs for mammalian cells[1, 22]) should allow for summation at relatively low frequencies, which was confirmed by experiments. Measured excitation thresholds for 340-ns pulses arranged in 5- and 100-pulse bursts with repetition rates from 1 Hz to 2 MHz are presented in Fig 1A. There is no summation at rates up to about 7 kHz, so thresholds for the bursts simply equal the threshold for a single 340-ns pulse. Summation at rates above 10 kHz is manifested by threshold reduction which follows a power function and is faster for 100-pulse bursts, which allow for the transmembrane potential build-up from a larger number of smaller pulses. At 2 MHz, the threshold for 100-pulse bursts drops from 400-500 to 10 V/cm. The threshold for 5-pulse bursts drops to its theoretical minimum (1/5 of the threshold for a single pulse) already at 0.3 MHz, suggesting the lack of any appreciable discharge between pulses.

The membrane potential induced by an nsPEF burst should be determined by the time-average EF during the burst. This value can be calculated as the threshold EF times the duty cycle. Indeed, the threshold value of the time-average EF plotted against the burst duration was the same regardless of nsPEF duration (Fig. 1B). This rule allows for predicting the threshold for a single “long” pulse whose duration is equal to the burst duration. As illustrated in Fig. 1C, a burst of 1000, 200-ns pulses at 3.3 MHz (duty cycle 67%, burst duration 300 μs) decreased nerve excitation threshold from 360±4 V/cm (for a single 200-ns pulse) to 1.63±0.2 V/cm. The time-average EF during such burst was 1.63 × 0.67 = 1.1 V/cm. This was exactly the measured threshold for a single 300-μs pulse (1.04 ± 0.2 V/cm). Measured threshold for one 200-μs pulse (pulse duration equals total time “on” in the burst) was only marginally higher, 1.21 +/− 0.17 V/cm. However, below we show that this simple relation between the efficiency of nsPEF bursts and a single long pulse is not always valid.

3.2. Burst nsPEF stimulation and electroporation of mouse ventricular cardiomyocytes (VCM)

nsPEF shocks are a promising modality for defibrillation[2, 5, 17, 18], but their adoption is hampered by the need for high voltages and possible VCM damage. We tested if the high voltage requirement can be offset by applying high-rate bursts. Indeed, compressing 1000, 200-ns pulses into 3.33 MHz bursts enabled excitation at only 80-200 V/cm (Fig. 2A,C,D), which is 10-20 times lower than with a single 200-ns shock [17, 18]. Excitation by nsPEF bursts (Fig. 2A, C) was distinctly different from the persisting Ca²⁺ elevation and cell shrinkage following membrane damage by stronger EF (Fig. 2B).

Fig. 2. Excitation and electroporation of isolated ventricular cardiomyocytes with nsPEF bursts.

(A-D) Ca²⁺ activation (measured by Fluo-4 fluorescence) by 1000 of 200-ns, 3.33 MHz pulses at indicated EF. (A) and (B): Imaging of excitation and electroporation, respectively; bursts applied at the arrow. Images were taken at 250-ms intervals (A) or when indicated (B, seconds). Bar: 40 μm. (C) Ca²⁺ transients at increasing EF, in the same cell. Poor recovery after 400 V/cm denotes electroporation. (D) Separation of excitation and electroporation, mean±s.e., p<0.001, t-test. (E-G) Action potential thresholds measured with FluoVolt dye. (E) Effect of pulse number for bursts of 100-, 200-, or 400-ns pulses, all with 100-ns interval. Increasing the pulses number decreased the threshold similarly for all nsPEF durations (left). Thresholds plotted against the total time “on” did not depend on pulse duration (center). Bursts of shorter nsPEF excited VCM at lower time-average EF (right). Mean±s.e., n=6-10. (F) Same result for bursts of 1000 pulses, 50- to 600-ns duration. Interpulse intervals were varied from 0.09 to 4.8 μs. Mean ± s.e., n=25-30.

VCM excitation by a single nsPEF was damaging already at the excitation threshold, causing abnormal action potentials [18] and long-lasting Ca²⁺elevation [17]. On the contrary, repeated stimulation with MHz bursts caused no damage (Fig. 2C). First signs of electroporation were observed at 350-400 V/cm (Fig. 2B, D), i.e., 2-3 times above the excitation threshold (p<0.001), allowing for a large safety “window”.

With a fixed 100-ns interval between nsPEF, thresholds decreased as a power function for pulse widths from 100 to 400 ns (Fig. 2E, left). Threshold was determined by the total time “on” within bursts, whereas the individual nsPEF duration did not matter (Fig. 2E, center). The threshold time-average EF plotted against burst duration was smaller for shorter pulses (Fig. 2E, right), contrasting nerve excitation data in Fig. 1B.

This unexpected result was verified in a separate set of experiments where VCM were excited by bursts of 1000 pulses; pulse duration was varied from 50 to 600 ns, and the interpulse intervals were changed from 90 ns to 4.8 μs. Plotting the time-average threshold EF values against burst duration yielded significantly smaller values for shorter pulses (Fig. 2F), consistent with the previous experiment in VCM but contrasting nerve excitation. This contradiction has not been explained and is indicative of a specific effect of nsPEF, distinctly different from conventional pulses.

3.3. Electroporation of non-excitable cells by nsPEF bursts

Electropermeabilization was measured in CHO and HEK cells, which do not express any voltage-gated channels[24, 8, 9]. In Fig. 3A, a burst of 100, 400 ns pulses was delivered to CHO cells loaded with a Ca²⁺ indicator Fluo-4. For different EF strengths, we tested pulse repetition rates from 1 kHz to 1 MHz, and counted the number of cells which did and did not respond with a detectable Ca²⁺ rise. Each datapoint in Fig. 3A represents the responding fraction from a total of 20-30 cells from 3 or more independent experiments. The EF values between 0.96 and 1.47 kV/cm caused no Ca²⁺ response at low repetition rates, but became increasingly more efficient at 0.3 MHz and above. This frequency was about two orders of magnitude higher than for nerve stimulation (Fig. 1A) and was consistent with the estimated charging time constant of about 1 μs [22]. A higher EF of 1.83 kV/cm was already efficient in 70% of cells at 1 kHz; the increase of the responding fraction at just 10 and 100 kHz is not fully understood, and may be a result of better detection since Ca²⁺ transients also became stronger.

Fig. 3. Electroporation of CHO (A) and HEK (B-F) cells by nsPEF bursts evidenced by Ca²⁺ uptake (A, D-F) or YO-PRO-1 dye uptake (B,C).

(A) 100-pulse, 400-ns bursts become increasingly more efficient above 0.1-0.3 MHz. Labels indicate the EF, in kV/cm; 20-30 cells per datapoint. (B) Time course of YO-PRO-1 uptake evoked by 1000 of 500-ns, 0.64 kV/cm pulses (arrow) at indicated duty cycle. (C) The last datapoint from (B) plotted against the repetition rate. Note the lack of effect below 1 MHz. (D) Ca²⁺ transients evoked by same bursts (arrow). (E) Similar Ca²⁺ transients evoked by one 500-μs pulse at indicated EF. (F) Maximum amplitude of Ca²⁺ transients from panels (D) and (E) plotted against the EF for a single 500-μs pulse (left) and against the duty cycle for nsPEF bursts. Mean±s.e., n=11-27 for all panels. HEK cells were stimulated in a low-conductance solution, see text.

With a low EF of 0.64 kV/cm, a burst of 1000, 500-ns pulses caused no permeabilization of HEK cells to either Ca²⁺ ions or YO-PRO-1 below 0.8-1 MHz (Fig. 3B-D). Above this frequency, electropermeabilization steeply increased. The time course and shape of Ca²⁺ transients evoked by MHz bursts (Fig. 3D) and single long pulses (Fig. 3E) were similar, suggesting the same basic mechanism of electroporation. The magnitude of a MHz burst-induced Ca²⁺ response could be matched to that of a single “long” pulse whose duration was made equal to the total time “on” during the burst (not to the total burst duration), Fig. 3F. The matching condition to evoke the response was:

$${\mathrm{E}}_{\mathrm{lp}}=({\mathrm{E}}_{\mathrm{nsPEF}})\mathrm{x}(\text{duty cycle})+0.05,$$

where E_lp and E_nsPEF are, respectively, the EF values (kV/cm) produced by a long pulse (500 μs) and by nsPEF (500 ns) at cell location. Such connection suggested that nsPEF bursts and matched single pulses should have comparable physiological consequences, such as the reduction of viability in severely electroporated cells. However, experiments did not confirm this conjecture.

3.4. Viability of EL4 cells electroporated by nsPEF bursts

Viability experiments required nsPEF treatment of relatively large cell populations in electroporation cuvettes. nsPEF in bursts had a triangular shape, with 200 ns width at 50% height, and the applied voltage did not always fully drop to zero between pulses (Fig. 4A). Bursts of 1000 nsPEF were applied at two different peak amplitudes (190 and 500 V, translating into 1.9 and 5 kV in cell suspension), and at low or high repetition rates (100 Hz and 3 MHz). The EF of 1.9 kV/cm was below the electroporation threshold for 200-ns pulses applied at 100 Hz, so bursts of up to 4000 pulses did not reduce cell viability. Same pulses applied at 3 MHz were above the threshold and reduced viability about twofold for all tested numbers of pulses (Fig. 4B, p<0.001 compared to 100 Hz, t-test). At 5 kV/cm, much smaller numbers of pulses reduced viability for both 100 Hz and 3 MHz bursts, but 3 MHz bursts were significantly more efficient (Fig. 4C, p<0.001).

Fig. 4.

Viability of EL-4 cells 24 h after exposure to nsPEF bursts. (A) Shape of high-rate nsPEF delivered to electroporation cuvettes. Dashed line denotes the time-average amplitude. (B) Effect of pulse number for 100-Hz and 3-MHz bursts for a low EF of 1.9 kV/cm. (C) Same dependence for a high EF of 5 kV/cm. (D) nsPEF bursts at indicated parameters kill about 50% of cells, whereas a single pulse (whose duration equals burst duration and the amplitude equals the average of the burst) has smaller effect. Mean± s.e., n=5-6. * p<0.0001, t-test.

With irregular pulse shape, time-average voltage and EF during the burst could not be calculated by multiplying nsPEF amplitude by the duty cycle. Instead, we digitized voltages during a burst with 0.2-ns resolution and calculated their average, which equaled 108 V (1.08 kV/cm) when the peak voltage of nsPEF in the burst was 190 V (1.9 kV/cm). In a separate series of experiments (Fig. 4D), we compared side-by-side the viability of cells treated by nsPEF bursts (1000 pulses, 400 ns, 3 MHz, 1.9 kV/cm) and by single pulses whose duration equaled burst duration (333 μs) and the EF equaled the time-average value in the burst (1.08 kV/cm). Viability after nsPEF bursts was about 50%, consistent with the previous set (Fig. 3B), whereas an “equivalent” single pulse was significantly less efficient and reduced the viability just to 87.6±1.4%, p<0.0001. Thus, MHz bursts of nsPEF were more efficient than predicted by their time-average amplitude, potentially due to unknown nsPEF-specific effects.

In summary, we explored excitation and electroporation by nsPEF bursts with up to MHz repetition rates. We found that for diverse targets and endpoints, MHz compression always increased nsPEF efficiency and decreased the threshold. The efficiency of nsPEF bursts increased with the number of pulses per burst, their amplitude, and duty cycle. The efficiency could be significantly different from single long pulses whose duration and amplitude equaled the duration and the time-average amplitude of nsPEF bursts, respectively. MHz compression of nsPEF bursts is a new modality in electrostimulation, with specific effects and new capabilities which have yet to be explored.

• Efficacy of nanosecond pulsed electric field was tested up to MHz frequencies

• Temporal summation enabled low-field bioeffects above a critical frequency

• MHz bursts of nanosecond stimuli activated cells at 10-100 lower thresholds

• MHz compression separated stimulation from electroporation at higher voltages

Abbreviations:

CAP

compound action potential

EF

electric field

nsPEF

nanosecond pulsed electric field