Bioelectrochemistry 100 (2014) 69–79

Contents lists available at ScienceDirect

Bioelectrochemistry

j ourna l homepage: www.e lsev ie r .com/ locate /b ioe lechem

In-vitro bipolar nano- and microsecond electro-pulse bursts forirreversible electroporation therapies

Michael B. Sano a,⁎, Christopher B. Arena a, Matthew R. DeWitt a, Dieter Saur b, Rafael V. Davalos a

a Virginia Tech, Blacksburg, VA 24061, United Statesb Technische Universität München, München, Germany

Abbreviations: DI, deionized; H-FIRE, high frequency iirreversible electroporation; RPMI, Roswell ParkMemorialto-cytoplasm ratio.⁎ Corresponding author at: 330 ICTAS, Stanger Street

States. Tel.: +1 540 231 1979.E-mail address: Mikesano@stanford.edu (M.B. Sano).

http://dx.doi.org/10.1016/j.bioelechem.2014.07.0101567-5394/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 13 August 2013Received in revised form 18 July 2014Accepted 25 July 2014Available online 4 August 2014

Keywords:High frequencyCancerNon-thermalAblation

Under the influence of external electric fields, cells experience a rapid potential buildup across the cell membrane.Above a critical threshold of electric field strength, permanent cell damage can occur, resulting in cell death. Typicalinvestigations of electroporation effects focus on two distinct regimes. The first uses sub-microsecond duration, highfield strength pulses while the second uses longer (50 μs+) duration, but lower field strength pulses. Here we in-vestigate the effects of pulses between these two extremes. The charging behavior of the cellmembrane andnuclearenvelope is evaluated numerically in response to bipolar pulses between 250 ns and 50 μs. Typicalirreversible electroporation protocols expose cells to 90monopolar pulses, each 100 μs in duration with a 1 secondinter-pulse delay. Here, we replace each monopolar waveform with a burst of alternating polarity pulses, whilekeeping the total energized time (100 μs), burst number (80), and inter-burst delay (1 s) the same. We show thatthese bursts result in instantaneous and delayed cell deathmechanisms and that there exists an inverse relationshipbetween pulse-width and toxicity despite the delivery of equal quantities of energy. At 1500 V/cm only treatmentswith bursts containing 50 μs pulses (2×) resulted in viability below10%. At 4000 V/cm, burstswith 1 μs (100×), 2 μs(50×), 5 μs (20×), 10 μs (10×), and50 μs (2×) duration pulses reduced viability below10%while burstswith 500ns(200×) and 250 ns (400×) pulses resulted in viabilities of 31% and 92%, respectively.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

When cells are exposed to electric fields there is a resultant structuralrearrangement of molecules in the lipid-bilayer that occurs due toincreases in transmembrane potential. At low field intensities, nano-scale pores form in the cell membrane temporarily increasing thetransport of molecules across the lipid-bilayer [1,2]. At high fieldintensities, the cell is unable to recover from the pore formation process,resulting in physical cell death [3,4]. This non-thermal mechanism hasbeen adapted clinically as a focal ablation technique knownas irreversibleelectroporation (IRE). IRE has shown to be safe and effective for treating avariety of cancerous pathologies [5–8], including tumors located in closeproximity to major blood vessels that can rapidly remove heat out ofthe ablation zone, rendering thermally-mediated therapies ineffective[9,10]. Additionally, due to the targeted disruption of the plasma mem-brane, the treatment volume is visible in real-time on multiple imagingplatforms [11].

rreversible electroporation; IRE,Institute medium; NCR, nucleus-

, Blacksburg, VA 24061, United

When IRE is employed in-vivo the electric field is typically pulsed fordurations of 50–100 μs to minimize joule heating effects which candamage the extracellular matrix. A typical treatment protocol involvesdelivering one pulse per second for 90 s [12,13] through needleelectrodes with a voltage to distance ratio of 1000–2000 V/cm [5,7,8].The electric field intensity required to induce IRE varies with tissue type,temperature, anisotropy and conductivity, cell size, as well as a numberof other physiological factors, but is typically in the vicinity of 500 V/cm.In-vitro, this field intensity changes significantly between cells in suspen-sion and those grown in a 3-D matrix and may be due to morphologicalchanges that occur when they attach to the extracellular matrix [14].

The mechanisms of cell death due to pulsed electric field have beenshown to vary with pulse length and amplitude. Microsecond andgreater duration pulses typically result in immediate cell death dueto irrecoverable damage to the cell membrane. In contrast, sub-microsecond pulses typically lead to the induction of apoptotic celldeath mechanisms [15,16] through caspase activation [17], calciumrelease from organelles [18], phosphatidylserine externalization[19], dissipation of mitochondrial membrane potential [20], andDNA damage [21]. An in depth analysis of these mechanisms can befound in [22]. Below 50 kV/cm, sub-microsecond pulses have beenfound to induce caspase-dependent apoptotic responses, while pulsesat higher field intensities induce a non-caspase dependent cascade [22].DNA damage due to sub-microsecond pulses has been demonstrated

Fig. 1.H-FIRE experimental setup: (A) 100 µL of cell suspensionwas added to a 2mmelec-troporation cuvette. [Inset] Mesh used to simulate the cell membrane and nuclearenvelope. (B) Schematic of the experimental burst containing a cycling of positive andnegative polarity pulses.

70 M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

in vitro [23] and in vivo [24], though the role of this damage in theapoptotic cascade is not well understood.

The amplitude of the appliedfield required to induce electroporationeffects has an inverse relation with pulse duration. Pucihar et al. foundthat the electric field required to electroporate seventy percent of cellsin-vitro, using a single pulse, increased from approximately 400 V/cmto 10 kV/cm as the pulse duration was reduced from 1 ms to 150 ns,respectively [25]. In-vivo, Nuccitelli et al. found that 100 pulses withduration of 300 ns were ineffective at reducing the size of melanomaswhen 10 kV/cm was applied. However, tumor volume reduced by 75%after eight days when the electric field intensity was increased to20 kV/cm for 200 pulses of the same duration [26]. Interestingly, forvery short duration pulses (60 ns), cell viability has been reported tobe affectedbymedia composition [27] indicating thatmolecular transportplays a significant role in the resulting cell death mechanisms. However,there is a clear inverse correlation between pulse length and the requireddose [28] and protocols with higher pulse numbers require lower electricfield intensities to induce cell death [29].

Bridging the gap between ultra-short and long duration pulses,Arena et al. recently showed that bursts of bipolar square waves withconstitutive pulses of 1 and 2 μs can be used to ablate brain tissue[white/gray matter] without inducing muscle contractions [30]. In thehigh frequency irreversible electroporation (H-FIRE) protocol presentedby Arena et al., bursts consisting of 50 bipolar square waves 2 μs induration per phase did not induce muscle contractions when delivereddirectly into the motor cortex with a voltage to distance ratio of4000 V/cm. In contrast, single monopolar pulses 200 μs in durationresulted in measurable muscle contractions with voltage to distanceratios as low as 500 V/cm. These bursts of short duration pulses havebeen theoretically shown to short through epithelial layers and producemore uniform treatment regions through heterogeneous tissues [31].

The ability to simultaneously achieve more predictable lesions inelectrically complicated tissues without inducing muscle contractionsby using bipolar bursts has important clinical implications. Lesionpredictability directly influences treatment outcomes and is requiredto ensure adequate tumor coverage with a lethal electric field whileminimizing damage to the surrounding healthy tissue. Eliminatingmuscle contractions obviates the need for neuromuscular blockade,which subsequently requires general anesthesia and monitoring ofrespiratory function. Thus, there is the potential to perform treatmentsoutside of the operating room. The burst characteristics required toachieve electroporationwithin this intermediate range of pulse durationsspanning 1 μs to 100 μs are still relatively unexplored [32], which serve asthe basis of this paper.

Here we present the in-vitro effects of high frequency bipolar bursts,shown in Fig. 1. Individual pulses within the burst are separated by 2 μsand sequential pulses alternate in polarity. The bursts are repeated onceper second for 80 s and each burst exposes cells to the appliedvoltages for 100 μs. This specific waveform serves three purposes whichare focused on clinical applications. First, our bursts contain an equivalent100 μs energized time to the monopolar pulses employed in clinicalirreversible electroporation systems, allowing for direct comparisonbetween our protocols with that of clinically viable electroporationsystems [5,8,33]. Secondly, these bursts are delivered once per secondto correspond to the approximate delivery rate of clinical systemswhich are synchronized with the patient's heartbeat to minimize risksof tachycardia. Last, it has been shown that these waveforms effectivelyeliminate muscle contractions associated with irreversible electro-poration therapies [30].

To demonstrate the effects of these pulses on the cell membrane andintracellular organelles, we present a finite element model of a cellincluding a nuclear envelope. The charging behavior of the cellularmembrane and nuclear envelope is evaluated in response to pulsesbetween 250 ns and 50 μs. A parametric analysis is conducted on theintra- and extracellular conductivity, nucleus-to-cytoplasm ratio, andpulse-to-pulse delay time. In-vitro experiments are presented to confirm

the non-thermal nature of the protocol and demonstrate irreversibleelectroporation within this intermediate pulse-width range. Thesefindings agree with previous work that shows there exists an inversecorrelation between pulse-width and toxicity, and shows that thisphenomenon remains despite delivery of equal quantities of energywithin each burst.

2. Methods

2.1. Numerical modeling

A numerical model of a cell in suspension was created in COMSOL4.2 using an impedance boundary condition scheme [34]. The solutiondomain consisted of a three dimensional cube with edge-lengths of0.1 mm. At the center of this domain, two spheres were createdrepresenting the cytoplasm and nucleoplasm. Within the solutiondomain, the Electric Currents module was used to solve for followingequations

∇ � J ¼ 0 A=m3� �

ð1Þ

J ¼ σ þ ε0εr∂∂t

� �E = A=m2

� �ð2Þ

E ¼ −∇U = V=mð Þ ð3Þ

71M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

whereU is the electric potential,E is the electricfield, and J is the currentdensity. One boundary was assigned as time dependent electrical po-tential

U ¼ U tð Þ = V: ð4Þ

The opposing boundary was assigned as the relative ground

U ¼ 0 V: ð5Þ

The remaining boundaries were defined as electrical insulation

n � J ¼ 0 A=mð Þ ð6Þ

where n is the normal vector to the surface, and J is the electrical currentdensity.

For each domain (media, cytoplasm, nucleoplasm), a separate ElectricCurrents physics module was used and the dependent electric potentialvariables Umedia, Ucyto, and Unuc for the media, cytoplasm, and nucleo-plasm domains were defined, respectively. These variables were thendefined to calculate the voltage across the cell membrane (Um) andnuclear envelope (Un)

Um ¼ Umedia−Ucyto = V ð7Þ

Un ¼ Ucyto−Unuc = V: ð8Þ

In each Electric Currents module, the boundaries representingmembranes were defined as impedance boundary conditions withreference potentials prescribed as the electric potential in the adjacent(Uref) domain

n � J1− J2ð Þ ¼ 1d

σ U−Urefð Þ þ ε0εm∂∂t U−Urefð Þ

� �A=m2

� �ð9Þ

where σ is the conductivity, ε0 is the permittivity of free space, εm is therelative permittivity, and d is the thickness of the cell membrane ornuclear envelope. For example, in the Media domain, the boundaryrepresenting the cell membranewas defined as an impedance boundarywith reference potential of Ucyto. In the Cytoplasm domain, the sameboundary representing the cell membranewas defined as an impedanceboundarywith a reference potential of Umedia. The boundarywas defined

Table 1Parameters used in numerical analysis.

Parameter Value

ε0/(m−3 kg−1 s4 A2) 8.85 × 10−12

σm/(S1 m−1) 0.2εm 80ε0dmem/m 5 × 10−9 [47]rc/m 6.55 × 10−6

σmem/(S1 m−1) 3 × 10−7 [48]εmem 8.57ε0 [70]σc/(S1 m−1) 0.3 [52]εc 154.4ε0 [49]NCR 0.8 [52]dne/m 40 × 10−9 [52]σne/(S1 m−1) 6 × 10−3 [52]εne 28ε0 [52]σnp/(S1 m−1) 1.35 [52]εnp 52ε0 [52]

Permittivity of free space (ε0), media conductivity (σm), media permittivity (εm), cellmembrane thickness (dmem), cell radius (rc), cell membrane conductivity (σmem), cellmembrane permittivity (εmem), cytoplasm conductivity (σc), cytoplasm permittivity(εc), nucleus-to-cytoplasm ratio (NCR), nuclear envelope thickness (dne), nuclearenvelope conductivity (σne), nuclear envelope permittivity (εne), nucleoplasm conductivity(σnp), nucleoplasm permittivity (εnp).

as a ‘thin layer’ and the electrical conductivity, relative permittivity, andsurface thicknesswere definedusing the values presented in Table 1. Thenuclear envelope consists of two individual lipid membranes separatedby the perinuclear space. To limit the complexity of the model andavoid improperly assessing the electrical properties of these individualcomponents, which are not readily available in the literature, we electedto lump these biological features into a single 40 nm membrane forwhich electrical properties representing their combined features areavailable.

The mesh was defined as a single free tetrahedral group with theelements between 1.8 and10 μmonedge, resulting in 19,353 tetrahedralelements. In a preliminary study of this model, finer and courser mesheswere used. Simulation times were more than doubled betweensuccessive refinements. The average deviation between the meshpresented here and the next successive refinement was less than2.0% and 5.5% for the cell membrane and nuclear envelop potentials,respectively. For eachparameter, solutionswere found in approximately22 min on a quad core 3.0 GHz processor with 8 GB of RAM. Results ofthe numerical simulations, using the values in Table 1, were comparedto those found using the analytical method presented by Kotnik andMiklavcic [35]. When calculating the maximum/minimum potentialsacross the cell membrane and nuclear envelop, the error between thenumerical and analytical solution was 0.15%/0.15% and 1.97%/0.89%,respectively. The values reported for Um and Un were derived from thegeometric boundary point closest to the energized electrode (θ = 0). Abackward differentiation formula (BDF) time stepping scheme wasused for all simulations. The solver was allowed to freely define thetime steps using an initial time step of 1 ns and a maximum time stepof 1 μs. The simulation was solved for a duration equal to three timesthe pulse width plus the inter-pulse delay.

2.2. Cell preparation and experimentation

In all experiments, cells were suspended in a buffer consisting of a5.5:1 ratio of culture media to low conductivity sucrose buffer (85 gsucrose, 3.0 g glucose, 7.25 mL RPMI, and 992.75 mL DI water) [36].The electrical conductivity of the cell suspension was measured with aconductivity meter prior to experimentation (Horiba B-173, Cole-Parmer, Vernon Hills, IL) to ensure a final conductivity of 0.2 S/m.Clark et al. reported that the conductivity of pancreatic tissue variedbetween 0.097 and 0.44 S/m for frequencies between 1 kHz and2MHz, respectively [37]. A media conductivity of 0.2 S/mwas chosentominimize the current delivered through the samplewhilemaintaininga conductivity valuewithin the range of those found in in-vivo tissue. Dueto limitations in our pulse generation system, higher conductivity bufferswould drive the pulse delivery systemoutside of its safe operating region.

PPT8182murine primary pancreatic tumor cells [38]were used in allexperiments. These cells have been shown to replicate humanpancreaticcancer in terms of histology, metastasis, and genetic alterations [38–41].Cells were cultured in DMEM (supplemented with L-glutamine, ATCC,Manassas, VA) containing 10% fetal bovine serum (Sigma Aldrich, St.Louis, MO) and 1% stock solution of penicillin/streptomycin (Invitrogen,Carlsbad, CA) at 37 °C in 5% CO2 in a humidified atmosphere. All cellswere harvested for experiments by trypsinization at 80% confluence.Suspensionswere centrifuged twice and resuspended in an experimentalbuffer at a concentration of 5 × 106 cells/mL. 100 μL of cell suspensionwas injected into a 2 mm gap cuvette (Model 620, Harvard Apparatus,Holliston, MA) immediately prior to pulse delivery. A schematic of theexperimental setup is shown in Fig. 1A.

The protocol for all experiments used the waveform presentedin Fig. 1B. The schematic depicts an example burst which contains arepeated sequence of individual pulses. The burst beginswith a positivepolarity pulse followed by a 2 μs pause, then a negative polarity pulsefollowed by another 2 μs pause. This cycling is immediately repeateduntil the voltage has been delivered for a total of 100 μs (50 μs in eachpolarity). Eighty bursts were delivered with a frequency of 1 Hz. Within

Fig. 2. Experimental waveforms: Applied voltage, U/kV, is presented as a function of time.t/μs. Each burst has a total on time of 100 μs, with 50 μs energized in each polarity. Represen-tative segments from bursts with (A) 250 ns, (B) 1 μs, and (C) 5 μs constitutive pulses.

72 M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

each burst, individual pulses had a single duration of 250 ns, 500 ns, 1, 2,5, 10, or 50 μs and therefore bursts contained 400, 200, 100, 50, 20, 10, or2 pulses, respectively to result in equivalent energized time. The 2 μsdelay timewas programed between sequential opposite polarity pulsesto protect the electronics from over-voltages due to ringing. Represen-tative examples of the bursts are shown in Fig. 2. The cells were exposedto electric potentials with voltage-to-distance ratios (E) of 1500, 3000,and 4000 V/cm. The temperature change in the cell suspension due topulsing was measured using fiber optic temperature probes (LuxtronFOT Lab Kit, LumaSense Technologies, Santa Clara, CA) inserted directlyinto the cell suspension.

For the in vitro studies, each of the treatment groups was repeated aminimum of three times (n= 3) and experiments for each group wereconducted on at least two different days. For each treatment, differentexperimental parameters, including sham exposure, were alternatedin a random sequence. After treatment, samples were split into twoequal 50 μL samples to be evaluated at 1 and 24 h time points. Thesamples were kept at room temperature for approximately 20–30 minprior to being placed on ice (1 hour group) or moved to the incubator(24 hour group) while the remaining experimental groups werecompleted. Approximately 1 h post exposure, viability was assessedusing a trypan blue exclusion assay. Cells which had been irreversiblyelectroporated were unable to exclude the dye and were stainedblue. Cells were counted visually using a hemocytometer and thepercentage viability was determined as

Viability1 hour ¼Nlive

Ntotal� 100 = % ð10Þ

rviabilty−1 hour ¼Viability1 hour−treatment

Viability1 hour−controlð11Þ

The average viability of sham control samples in the 1 hour timegroupwas greater than 85%. Samples to be analyzed at 24 hwere placedin separate wells in a 12-well pate containing a total of 1 mL of culturemedia and maintained at room temperature until the well plate wasfull (approximately 30 min). At this point the well plate was placed inan incubator at 37 °C and 5% CO2 for 24 h. Viability was then assessedusing an Alamar blue metabolism assay (Life Technologies, GrandIsland, NY) using the manufacturer's recommended procedure. Briefly,100 μL/mL stock Alamar blue solution was added to each well. After4 h, the samples were read using a spectrophotometer at 570/600 nmwavelengths. For each sample, the absorbance was measured in threeseparate wells and averaged. Additional measurements were takenfor sample media without cells and for control cell samples whichwere not exposed to an electric field. The percentage viability wasdetermined as

rviability− 24 hour ¼Isample−Imedia

Icontrol−Imediað12Þ

where I is the relative intensity measurement from the spectropho-tometer. In general, trypan blue analysis and metabolism assayscomplement each other quite well. Ibey et al. previously showedthat metabolism assays mirrored those from trypan blue analysisafter nano-second pulsed electric field exposure [28]. The Alamarblue assay used in this study is well established for measuringcytotoxicity inmammalian cells [42]. Reduction rates for cells seededbetween 2.5 × 103 and 2 × 106 cells/mL were measured to ensurethat the sham population did not completely reduce the Alamarblue solution (results not shown) and a 4 hour incubation timewith 2.5 × 105 cells/mL was determined to be optimal. Viabilitydata for both the 1 hour and 24 hour groups were normalized tothe sham control groups. Statistical analysis of the data was completedusing JMP Pro V. 10.0 (SAS Institute Inc., Cary, NC).

2.3. Electronics

Waveforms were generated using an arbitrary function generator(AFG3021C, Tektronix Inc., Beaverton, Oregon), which were amplifiedby a custom built high voltage pulse generator capable of +/− 1000 Voutputs through high impedance loads (Applied Energetics, Tucson,AZ, USA). Output waveforms were visualized using an oscilloscope(DPO2002B, Tektronix Inc., Beaverton, Oregon) after the voltage wasattenuatedusing a 50MHz1000×high voltage probe (P5210A, TektronixInc., Beaverton, Oregon) and the current was measured using an activeclamp on 50 MHz current probe (TCP305, Tektronix Inc., Beaverton,Oregon). Short circuit protection resistors on the output limited ourmaximum output voltage through the 2 mm cuvettes to approximately800 V (4000 V/cm).

3. Results and discussion

3.1. Numerical modeling

As shown in Fig. 3, under the influence of a 1500 V/cm electric field,the potential drop across the cell membrane (Um) and nuclear envelope(Un) reaches maximums of 1.47 V and 0.28 V, respectively. Um reaches50% of the maximum value in 0.34 μs, 70.7% in 1.11 μs, and 99.99%maximum in 7.92 μs. Un reaches 99.99% max in 145 ns and falls backbelow 70 mV in approximately 0.94 μs. This brief charging and

Fig. 3. Finite element simulations: The applied electric field, E/(kV/cm), voltage dropacross the cell membrane, Um/V, and nuclear envelop, Un/V, are presented as a functionof time, t/μs. (A) A bipolar squarewavewith 10 ns rise and fall timeswas used to simulatethe maximum (B) transmembrane potential of the cell membrane (Um) and (C) nuclearenvelope (Un). Note thedoublepeak inUn that occurs as thefirst pulse falls to zero followedby a 2 μs delay and onset of the negative polarity pulse.

Fig. 4. Parametric analysis of experimental parameters: The voltage drop across the cellmembrane, Um/V, and nuclear envelop, Un/V, are presented as a function of time, t/μs.(A) Pulse width, (B) media conductivity, (C) pulse-to-pulse delay time. Dashed linesrepresent the transmembrane potential of the cellmembrane (Um) and solid lines representthe transmembrane potential of the nuclear envelope (Un). Note that the axes for Um andUn

have different scales.

73M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

discharging of the nuclear envelop is due to current that flowswithinthe cytoplasm as the cell membrane is charging. This transientcurrent increases the potential across membranes surrounding thenucleus and organelles. These intracellular components are smallerthan the cell and their exposure to currents is brief resulting in asmaller potential increase.

As the positive polarity pulse falls, the cell membrane begins todischarge resulting in a second current flow within the cytoplasmin the opposite direction, as compared to the rising pulse edge. Thisresults in the formation of a negative potential across the nuclearenvelope. This negative potential reaches a minimum of −0.28 Vand falls below −70 mV in a similar 0.94 μs. The rising edge of thenegative polarity pulse creates a similar decrease in Un creating aninteresting double peak in the membrane potential of the nuclearenvelope. This second peak reaches a value of −0.29 V. Thoughthis peak is only 10 mV different than the maximum achieved bythe initial pulse, it suggests that optimization of the pulse lengthand delay time between pulses could result in an increased effecton intracellular membranes.

In this manuscript, we elected to disregard the effects of electro-poration on the cell membrane to simplify our analysis. However,in the case of electroporation, current would be allowed to flowthrough the cytoplasm and a sustained potential would be inducedacross the intracellular membranes, thereby reducing the first negativepeak in Un.

3.2. Analysis of experimental parameters

Fig. 4 presents a parametric analysis of variables which can becontrolled experimentally. The pulse duration, shown in Fig. 4A, directlyimpacts themaximumUmachieved and theduration thatUm is elevatedabove the 1 V critical threshold. Pulses that are shorter than 1 μs do notelevate the Um above this threshold. As pulse duration increases beyond1 μs, Um saturates to amaximum value of 1.47 V. In contrast, because Un

rises rapidly in comparison to theUm, the effects on the nuclear envelopeareminimally impacted by the pulse duration. Regardless of pulsewidth,the Un reaches a maximum value within 145 ns. For pulses 1 μs or less,

74 M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

the Un does not completely return to zero before the falling edge of thepositive pulse, muting the negative Un response.

It has been observed that pore formation behavior occurswithin 1 μsafter Um is elevated above 1 V, quenching further increases in potential[43], after which new pore formation is limited and pore expansiontakes over as the dominant phenomena [44,45]. At the field strengthspresented here, pulses 1 μs in duration and shorter may not efficientlyresult in pore expansion within the cell membrane [46].

The conductivity of the sample media, Fig. 4B, contributes signif-icantly to the charge–discharge behavior of the cell membrane andthe nuclear envelope. At low media conductivities (0.01 S/m), themedia presents a significant resistance to current flow and the cellmembrane charges slowly. Similarly, this low conductivity mediaminimizes the current which can flow through the cytoplasm,muting the maximum Un achieved. As the media conductivityincreases, the cell membrane charges more quickly, saturating asthe conductivity is increased above 1 S/m. Based on these simulations,a media conductivity of 0.2 S/m used experimentally is a compromisebetween membrane charging times and current output required fromthe pulse generator. Increasing media conductivity may have resultedin slightly faster membrane charging times.

The delay between positive and negative polarity pulses, Fig. 4C, hasa negligible effect on the transmembrane potential (Um); though, it hasa significant impact on the nuclear envelope (Un). The falling edge ofthe positive pulse results in a negative potential build-up on thenuclear envelope. Un reaches a relative maximum approximately140 ns after the falling edge of each pulse. For long delays betweenpulses, this potential decays back to zero. In contrast, as the delayis contracted, Un is compounded by the rising edge of the negativepolarity pulse. Ultimately, as the delay is decreased to 140 ns orless, an effective doubling of the Un is achieved.

Typically, driver circuits for high voltage solid state switchesemploy a dead time between changes in polarity to avoid shoot-through (short circuiting the power supply to ground) or to protectthe electronics from deleterious effects caused by ringing. Thisdelay between changes in polarity is highly dependent on the topologyof the pulse generation circuit and is typically between 100 and 500 ns.Based on these simulations, burstswith a 100 ns delay between changesin pulse polarity will continue to achieve a doubling of the potentialacross the nuclear envelop. An additional requirement to achieve thisdoubling in Un is that the potential across the nuclear envelope mustbe allowed to decay back to zero before the applied voltage is turnedoff. In this scenario, all pulses which are 0.94 μs in duration or longerresulted in approximately a 2× increase in Un versus the single pulsemaximum.

The role of DNA damage in the pulsed electric field (PEF) apoptoticcascade is not fully understood and the nucleus is not typically thetarget for PEF therapy. However, intrinsic and extrinsic apoptoticcell death processes are associated with field strength dependenteffects on mitochondria and the endoplasmic reticulum. If waveformoptimization can be used to double the increase in the transmem-brane potential of these organelles, as shown in Fig. 4C, then loweramplitude electric fields would be needed to induce the associatedapoptotic cascades. Alternatively, by finely tuning the pulse widthsand inter-pulse delays it may be possible to enhance DNA damageprocesses allowing for further study of this mechanism in the PEFapoptotic cascade. Unfortunately, experimental investigation of wellcontrolled 100–500 ns inter-pulse delay scenarios was inhibited byringing in the output voltages of our current system and is left as thesubject of future work.

Fig. 5.Cell property parametric analysis: The voltage drop across the cellmembrane, Um/V,andnuclear envelop, Un/V, are presented as a function of time, t/μs. (A)Nucleus–cytoplasmratio, (B) cytoplasm conductivity, and (C) cell membrane permittivity. (D–E) The voltagedrop across the cell membrane, Um/V, and nuclear envelop, Un/V, for a model of benignand cancerous cells. Note that the axes for Um and Un have different scales.

75M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

3.3. Analysis of cell electrical properties

Electrical properties for the cell membrane, nuclear envelope,cytoplasm, and nucleoplasm are readily available in the literature[47–52]. Subuncu et al. report a cytoplasmic conductivity between0.3 and 0.6 S/m [53]. Labeed et al. report increases in conductivity

Fig. 6. Simulation of membrane potentials due to 250 ns and 1 μs experimental pulses: Theapplied electric field, E/(kV/cm), voltage drop across the cell membrane, Um/V, and nuclearenvelop, Un/V, are presented as a function of time, t/μs. (A) 1500 kV/cm 250 ns impulseand (B) the resulting transmembrane potential of the cell membrane (Um) and nuclearenvelope (Un). (C) 1500 kV/cm 1 μs impulse and (D) the resulting transmembrane potentialof the cellmembrane (Um) andnuclear envelope (Un). Dashed lines represent the transmem-brane potential of the cell membrane (Um) and solid lines represent the transmembranepotential of the nuclear envelope (Un).

from 0.28 S/m to 0.45 S/m as cells begin to undergo apoptosis [54].Ron et al. report a conductivity of 0.724 S/m and 0.93 S/m for pre-osteoblast cells and normal canine kidney cells, respectively [55].Mulhall et al. found cytoplasm conductivities of 0.71, 0.42, 0.26, and0.25 S/m for normal keratinocytes, abnormal keratinocytes, and fortwo different malignant keratinocytes, respectively [56] . Additionally,Chen et al. show that drug resistant cells have a lower cytoplasmicconductivity than non-drug resistant cells [57]. These results provideevidence of decreasing cytoplasmic conductivity with cells transitionfrom benign to malignant.

Yuan et al. show an increase in nucleus-to-cytoplasm (NCR) ratiofrom 0.45 to 0.49 and from 0.40 to 0.49 as cancer cells achieve drugresistance. Similarly, Helczynska et al. show histologically, that theNCR increases from 0.3 to 0.8 as a function of tumor grade, with higherNCRs for increasingly malignant cancers [58]. Salmanzadeh et al.showed that the specific membrane capacitance of a syngeneic cellline increased from 15.39 mF/m2 to 26.42 mF/m2 as the cells becamesuccessively more malignant [59]. This translates into an increase inrelative membrane permittivity from 8.70 to 14.92.

A parametric analysiswas conducted using cytoplasmic conductivityvalues of 0.7, 0.475, and 0.25 S/m, an NCR of 0.3, 0.55, and 0.8, and amembrane permittivity of 9, 12, and 15 to represent this transitionfrom benign to intermediate to metastatic, respectively. We modeledthe response of a ‘benign’ cell having cytoplasmic conductivity of0.7 S/m, NCR of 0.3, and membrane permittivity of 8.7. A ‘metastatic’cell was modeled as having cytoplasmic conductivity of 0.25 S/m,NCR of 0.8, and a membrane permittivity of 15. All other values(Table 1) were held constant.

A parametric analysis was conducted using an NCR of 0.3, 0.55, and0.8, cytoplasmic conductivity values of 0.7, 0.475, and 0.25 S/m, and amembrane permittivity of 9, 12, and 15 to represent this transitionfrom benign to intermediate to metastatic, respectively. The NCR,Fig. 5A, has a negligible effect onUmandnotable effect onUn. As expectedfrom electromagnetic theory [60], the potential across an the nuclearenvelope is related to the equation

ΔU ¼ 1:5rEcosθ = V ð13Þ

where r is the radius of the nucleus and E is the electric field towhich thecell is exposed to. However, other dielectric properties of the nucleusmay affect the membrane charging time [35,61]. As the NCR increasesin Fig. 5A, Un also increases. The cytoplasm conductivity, Fig. 5B, has anegligible impact on themaximum amplitude of Um and Un. The permit-tivity of the cell membrane, Fig. 5C, impacts the charge and discharge ofthe cell membrane and the nuclear envelope. A higher permittivitycauses the Um to increase slightly slower than the lower permittivitycells. This slower charging time of the cell membrane results in thenuclear envelope reaching a slightly higher transmembrane potential.

From the numerical simulations, it is anticipated that cells with alarger NCR will achieve higher Un amplitudes than cells of similar sizewith smaller NCR. A high NCR has been associated with the aggressive-ness of malignant cells and is used as a parameter in grading cancers[65–68]. Additionally, it has been shown that an increase in invasivenessand metastatic potential has been correlated to cell membrane ruffling,which leads to highermembrane capacitances in aggressive cells [59,69,70]. We therefore modeled the response of a ‘benign’ cell havingcytoplasmic conductivity of 0.7 S/m, NCR of 0.3, andmembrane permit-tivity of 8.7. A ‘metastatic’ cell was modeled as having cytoplasmicconductivity of 0.25 S/m, NCR of 0.8, and a membrane permittivity of15. All other values (Table 1) were held constant.

In numerical simulations (Fig. 5D), a normal cell model experiencesa |Un| ≈ 0.14 V while a cancer cell model reaches |Un| ≈ 0.32 V. Thenucleus in the cancer cell model reaches a potential approximately2 times higher than the normal cell model as a result of changes inNCR. This effect is amplified further if the delay between pulses isreduced to 100 ns (Fig. 5E) where |Un| ≈ 0.6 V for the cancer cell

Fig. 7. Change in media temperature during exposure to 4000 V/cm: The change intemperature,ΔT/K, is presented as a function of time, t/s. Burstswith 50 μs and 250ns con-stitutive pulses resulted in similar temperature rises. ΔT = T − Tref for each experiment,where Tref = 23.45 ± 0.55 °C.

Fig. 8. Cell death occurs due to immediate and delayed mechanisms: The relativeviability, rviability, is presented as a function of pulse width, Δtp/μs. Relative viabilityof cells 1 and 24 h after exposure to (A) 1500 V/cm, (B) 3000 V/cm, (C) 4000 V/cmbursts. In all experiments cells were exposed to 80 bursts each with an energizedtime of 100 μs. Error bars represent the standard deviation after a minimum ofthree (n = 3) randomized experiments. Stars (*) denote statistical significancebetween 1 and 24 hour time points (α ≤ 0.1).

76 M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

model. It is anticipated thatmalignant cells will experience an increasedresponse to bipolar pulses due the increase in cell membrane chargingtime, resulting from an increased membrane capacitance, coupledwith increased NCR ratio. However, future work will be required todetermine if these burst have an increased efficiency at targetingaggressive cells.

3.4. Numerical simulation of experimental pulses

The simulation results presented in Figs. 3 to 5 represent theresponse to a squarewavewith 10 ns rise and fall times. Experimentally,thewaveforms exhibited ringing effects on the rising edge and after thefalling edge as shown in Fig. 6A and C. Fig. 6B and D shows the celltransmembrane potential (Um) and nuclear transmembrane potential(Un) resulting from experimental 250 ns and 1 μs pulses, respectively.As in the square wave case, the falling edge of the pulses results inan increased Un in the opposite polarity. The ringing in the outputwaveform causes an additional minor increase in Un. At 1500 V/cmthe first rising edge of a 250 ns pulse results in an Un amplitudemaximum of 0.21 V. The falling edge and ringing of the same pulseresults in a maximum Un amplitude of 0.25 V, a 19% increase.

For a 1 μs experimental pulse, |Um| reaches a maximum of 1.24 Vwhile |Un| reaches a maximum of 0.32 V. The magnitude of Um for thisexperimental pulse is approximately equal as in the case of ideal squarewave, predicted in Fig. 4A (1.21 V). However, the magnitude of Un forthis experimental pulse (0.32 V) is greater than the value predictedin Fig. 4A (0.29 V). This is due to the ringing which occurs after theexperimental pulses fall back to zero.

As the pulse length increases, the initial Un response is allowed tofall back towards zero. The result is that for longer pulses, the negativegoing edge and subsequent ringing have an increased effect. For similarfield strengths, a 5 μs pulse results in Un amplitude change from 0.24 Vto 0.36 V, a 50% increase (not shown). For these cases, the peakamplitude of the ringing is 46–52% that of the pulse amplitude andlasts for less than 200 ns.

3.5. Experimental results

Experiments were conducted with an initial sample temperaturebetween 22 and 25 °C. At 4000 V/cm all experimental groups resultedin a temperature rise less than 3.5 °C. Representative temperatureprofiles for experiments with 50 μs and 250 ns constitutive pulses areshown in Fig. 7. The temperature increase for bursts with 250 ns pulsesis similar to the increase for longer duration pulses. This is likely due tothe delivery of an equivalent quantity of energy in each burst regardlessof the duration of the constituent pulses. The starting temperatureof the experiments ensured that the temperature never rose above

37 °C, mitigating the possibility of temperature as a confounding factor,affecting the viability of cells.

Fig. 8 shows the viability of the samples 1 and 24 h after treatmentfor field strengths of (Fig. 8A) 1500 V/cm, (Fig. 8B) 3000 V/cm, and(Fig. 8C) 4000 V/cm. There is a clear inverse relationship betweenconstituent pulse length and viability, with longer duration pulsesresulting in a lower viability for both the 1 and 24 hour viability studies.

Specifically, at 1500 V/cm, bursts containing 50 μs pulse (2×)resulted in a 1 hour post-treatment viability of 31% which reducedto 3% after 24 h. The 1500 V/cm bursts containing pulses between250 ns (400×) and 10 μs (10×) resulted in 1 hour viabilities above50% and notably, pulses 2 μs (5×) and shorter had viabilities of 85%or greater, similar to sham treatments. For this field strength, burstscontaining 10 μs pulses had the largest change in viability over 24 h,49%, while 250 and 500 ns pulses resulted in a negligible change inviability compared to controls. Significant changes in viabilityoccurred between the 1 and 24 hour time points for bursts withpulses 2 μs and longer. It is interesting that 10 and 50 μs pulsesresulted in delayed cell death, however, the mechanism of action isunclear.

Cell viabilitywas significantly lower for 3000V/cm versus1500V/cmbursts when the pulse duration was 1 μs or longer. After 24 h, the

77M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

viability for 2 to 50 μs pulses reduced to less than 5% at 3000 V/cm.Between 3000 V/cm and 4000 V/cm, the most significant impact onviability occurred for 500 ns pulses. For all field strengths, 250 ns pulseshave a minimal impact on cell viability.

For bursts containing 250 ns pulses, the difference in viability after1500, 3000, and 4000 V/cm treatments was not statistically significant(α≤ 0.1). All other pulse-widths had a statistically significant differencebetween the 1500 V/cm and 3000 V/cm treatments at each time point(α ≤ 0.06). Between the 3000 and 4000 V/cm treatments, 5 μs (1 h),500 ns (1 h), and 500 ns (24 h) groups had statistically differentviabilities (α ≤ 0.03)

Interestingly, this study shows that viability is not directly correlatedto the energy dose delivered. This conforms to the results presented byothers that electropermeabilization [60] and lethal [28] effects ofmonopolar pulses of different pulse widths exhibit a complex relation-ship that cannot be correlated to the quantity of energy deliveredalone. The inverse correlation between pulse length and toxicity pre-sented may be related to the cell membrane charging time, calculatedhere as between 1.11 and 7.92 μs. Fig. 9 shows the effect of multiplepulses within each burst on the time in which Um and Un are elevatedabove critical thresholds. A single cycle of 1500 V/cm 250 ns pulses,one positive and one negative, increases Um above 1 V for only 200ns total. However, the cumulative effect of the full burst (400 totalpulses) increases Um above 1 V for approximately 40 μs. At 1500 V/cm

Fig. 9.Bursts have cumulative effect on the timemembrane potentialswhich are above critical thcritical threshold is presented as a function of pulse width, Δtp/μs. [A, B, E, F, I, J]. Time for whicwhich the nuclear envelope has a potential drop (Un) greater than 0.5, 0.75, or 0.9 V. Valueswerpositive and one negative pulse] may increase the transmembrane potentials only briefly, hoduration that these cellular components experience an elevated potential.

(Fig. 9A–B) time above the 1 V threshold increases as constitutivepulse width increases. This protocol reaches a maximum of approxi-mately 99.8 μs for bursts with 50 μs constitutive pulse widths, whichonly have one cycle. At 3000 and 4000 V/cm, Fig. 9 (E–F, I–J), bursts ofshorter pulses elevate Um above 1 V for a longer duration than thosewith longer pulse durations. This is additional time is due to themultiple charge/discharge cycles of the cell membrane which occurduring a burst of shorter pulses, however, this appears to have anegligible impact on cell viability. Though not examined here, pulsesenergized for less than the membrane charging time may result inlimited pore expansion, minimizing lethal effects.

At 1500 V/cm, none of the pulse durations elevated Un above 0.5,0.75, or 0.9 V (Fig. 9C–D). The voltages 0.5, 0.75 and 0.9 V were chosenrather arbitrarily, to demonstrate the induced voltage on the nuclearenvelope for each set of pulse parameters used. 1.0 V threshold wasnot reached for any simulation. At 3000 V/cm (Fig. 9G–H), all pulsedurations are able to increase Un above the 0.5 V threshold. Thecumulative impact of a full burst results in Un increasing above 0.5 Vfor a substantially longer duration for shorter constitutive pulses. At4000 V/cm (Fig. 9K–L), some pulse durations are able to elevate Un

above 0.75 and 0.9 V. Interestingly, 500 ns pulses result in greatercumulative time above all of the voltages (0.5, 0.75, and 0.9 V) thanany other pulse durations. This may help explain why 500 ns burstsresulted in significant changes in viability between 1 and 24 h and

resholds: The time, t/μs, forwhich the cellmembrane or nuclear envelope is greater than ah the cell membrane has a potential drop (Um) greater than 1 V. [C, D, G, H, K, L]. Time fore calculated using an idealizedwaveformwith 10 ns rise and fall times. A single cycle [onewever, cumulative effect of having multiple pulses per burst substantially increases the

78 M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

250 ns did not. However, the exact mechanism cannot be determinedfrom this study.

For all bursts containing pulses 1 μs in duration or longer, the viabilityat 3000 V/cmafter 24 h is lower than the corresponding viability at 4000V/cm after 1 h. This has interesting implications for in-vivo applicationsas it indicates that ablation sizes may grow over time and that immedi-ate observation may be inadequate to predict the total volume treated.From the numerical simulations, it is anticipated that cells with a largercytoplasm–nucleus ratio will achieve higher Un amplitudes than cells ofsimilar sizewith a smaller ratio. A high nucleus–cytoplasmic ratio (NCR)has been associated with the aggressiveness of malignant cells and isused as a parameter in grading cancers [61–64]. Additionally, it hasbeen shown that an increase in invasiveness and metastatic potentialhas been correlated to cell membrane ruffling, which leads to highermembrane capacitances in aggressive cells [59,65,66].

4. Conclusion

We found, through finite element simulations, that by reducing thedelay between consecutive bipolar pulses, it would be possible toachieve a doubling of the nuclear transmembrane voltage, hence,possibly promoting intracellular effects induced by electric pulses.We also found that the NCR and cell membrane permittivity play asignificant role in the charging characteristics of the nuclear envelopeand that cancer cells may possess some properties which would resultin more profound intracellular effects as compared to normal cells.However, our simplified model has some limitations. Cells weremodeled as simple spheres to reflect the shape of the cells in theirnon-adhered state. Media composition, cell shape, adherence, tempera-ture, repetition rate, waveform shape, and cell cycle are known to alterthe response of cells to pulsed electric fields. In vivo, cells typically takeon more complex, elongated, or spindled shapes which can alter theeffects of pulsed electric fields on transmembrane potential. Cells intissue are affected by local inhomogeneity and the responses of cells intheir immediate vicinity which was not accounted for here.

NCR, cytoplasm conductivity, and cell membrane permittivity play asignificant role in the charging characteristics of the nuclear envelope.Experimentally we found that bursts of bipolar square waves increasedthe media temperature less than 3.5 °C when the total energized timeper burst was held constant at 100 μs and eighty bursts were delivered.The resulting cellular responses are therefore limited to those relateddirectly to non-thermal phenomena. For the bursts of bipolar pulsespresented, there exists an inverse correlation between pulse-widthand toxicity despite the delivery of equal quantities of energy in eachburst. The changes in cellular viability over the 24 hour post treatmentshow presence of both instantaneous and delayed cell death processes,however, the exact mechanisms are unknown.

To the best of our knowledge, this is the first experimental parametricanalysis on the effects of bipolar square wave bursts with pulsesbetween 0.25 and 50 μs. In the 3000 V/cm treatment groups, cellviability was reduced to 4.0%, 0.5%, 0.3%, and 1.0% for bursts containing2, 5, 10, and 50 μs pulses, respectively. In the 4000 V/cm treatmentgroups, cell viability was reduced to 3.8%, 1.4%, 0.9%, 0.8%, and 0.8% forbursts containing 1, 2, 5, 10 and 50 μs pulses, respectively. Rubinskyet al. [67] showed that ten 100 μs monopolar pulses at 2000 V/cmresulted in a viability of 70%. In the same study, they showed thatseventy-five 100 μsmonopolar pulses at 250 V/cm resulted in a viabilityof 10–20% while ninety 100 μs monopolar pulses at 250 V/cm reducedviability to 0–10%. Arena et al. [14] showed that after eighty 100 μsmonopolar pulses at 1500 V/cm, cell viability was approximately 8%and this protocols is consistent with those currently being employedsuccessfully in clinical applications of irreversible electroporation inthe prostate [68], pancreas [8], and liver [69]. The comparable level oftoxicity resulting from the bipolar burst protocol presented here indi-cates that it may be advantageous in in-vivo therapies where musclecontractions due to longer duration monopolar pulses are undesirable.

Acknowledgments

This material is based upon work supported in part by the NationalScience Foundation under Awards CAREER CBET-1055913 and IIP-1265105.

References

[1] H.J. Scheffer, K. Nielsen, M.C. de Jong, A.A. van Tilborg, J.M. Vieveen, A.R. Bouwman, S.Meijer, C. van Kuijk, P.M. van den Tol, M.R. Meijerink, Irreversible Electroporation forNonthermal Tumor Ablation in the Clinical Setting: A Systematic Review of Safetyand Efficacy, Journal of Vascular and Interventional Radiology 25 (7) (2014)997–1011.

[2] A. Zupanic, B. Kos, D. Miklavcic, Treatment planning of electroporation-basedmedical interventions: electrochemotherapy, gene electrotransfer and irreversibleelectroporation, Phys. Med. Biol. 57 (2012) 5425–5440.

[3] B. Rubinsky, Irreversible electroporation in medicine, Technol. Cancer Res. Treat. 6(2007) 255.

[4] R.V. Davalos, B. Rubinsky, Temperature considerations during irreversibleelectroporation, Int. J. Heat Mass Transf. 51 (2008) 5617–5622.

[5] R.E. Neal II, J.H. Rossmeisl Jr., P.A. Garcia, O.I. Lanz, N. Henao-Guerrero, R.V. Davalos,Successful treatment of a large soft tissue sarcoma with irreversible electroporation,J. Clin. Oncol. 29 (2011) E372–E377.

[6] P.A. Garcia, J.H. Rossmeisl Jr., R.E. Neal II, T.L. Ellis, J.D. Olson, N. Henao-Guerrero,J. Robertson, R.V. Davalos, Intracranial nonthermal irreversible electroporation:in vivo analysis, J. Membr. Biol. 236 (2010) 127–136.

[7] K.R. Thomson, W. Cheung, S.J. Ellis, D. Federman, H. Kavnoudias, D. Loader-Oliver, S. Roberts, P. Evans, C. Ball, A. Haydon, Investigation of the safety of irreversibleelectroporation in humans, J. Vasc. Interv. Radiol. 22 (2011) 611–621.

[8] R.C. Martin II, K. McFarland, S. Ellis, V. Velanovich, Irreversible electroporationtherapy in the management of locally advanced pancreatic adenocarcinoma, J.Am. Coll. Surg. 215 (2012) 361–369.

[9] R.V. Davalos, L.M. Mir, B. Rubinsky, Tissue ablation with irreversible electroporation,Ann. Biomed. Eng. 33 (2005) 223–231.

[10] O.O. Adeyanju, H.M. Al-Angari, A.V. Sahakian, The optimization of needle electrodenumber and placement for irreversible electroporation of hepatocellular carcinoma,Radiol. Oncol. 46 (2012) 126–135.

[11] L. Calmels, B. Al-Sakere, J.P. Ruaud, A. Leroy-Willig, L.M. Mir, In vivomri follow-up ofmurine tumors treated by electrochemotherapy and other electroporation-basedtreatments, Technol. Cancer Res. Treat. 11 (2012) 561–570.

[12] P.A. Garcia, J.H. Rossmeisl Jr., R.E. Neal II, T.L. Ellis, R.V. Davalos, A parametric studydelineating irreversible electroporation from thermal damage based on a minimallyinvasive intracranial procedure, Biomed. Eng. Online 10 (2011).

[13] M.B. Sano, R.E. Neal, P.A. Garcia, D. Gerber, J. Robertson, R.V. Davalos, Towards thecreation of decellularized organ constructs using irreversible electroporation andactive mechanical perfusion, Biomed. Eng. Online 9 (2010) 83.

[14] C.B. Arena, C.S. Szot, P.A. Garcia, M.N. Rylander, R.V. Davalos, A three-dimensionalin vitro tumor platform for modeling therapeutic irreversible electroporation,Biophys. J. 103 (2012) 2033–2042.

[15] S.J. Beebe, P.M. Fox, L.J. Rec, E.L.K. Willis, K.H. Schoenbach, Nanosecond, high-intensity pulsed electric fields induce apoptosis in human cells, FASEB J. 17(2003) 1493–1495.

[16] S.J. Beebe, P. Fox, L. Rec, K. Somers, R.H. Stark, K.H. Schoenbach, Nanosecond pulsedelectric field (nsPEF) effects on cells and tissues: apoptosis induction and tumorgrowth inhibition, IEEE Trans. Plasma Sci. 30 (2002) 286–292.

[17] P.T. Vernier, A. Li, L. Marcu, C.M. Craft, M.A. Gundersen, Ultrashort pulsed electricfields inducemembrane phospholipid translocation and caspase activation: differentialsensitivities of Jurkat T lymphoblasts and rat glioma C6 cells, IEEE Trans. Dielectr. Electr.Insul. 10 (2003) 795–809.

[18] J.A. White, P.F. Blackmore, K.H. Schoenbach, S.J. Beebe, Stimulation of capacitativecalcium entry in HL-60 cells by nanosecond pulsed electric fields, J. Biol. Chem.279 (2004) 22964–22972.

[19] S.J. Beebe, J. White, P.F. Blackmore, Y. Deng, K. Somers, K.H. Schoenbach, Diverseeffects of nanosecond pulsed electric fields on cells and tissues, DNA Cell Biol. 22(2003) 785–796.

[20] W. Ren, S.J. Beebe, An apoptosis targeted stimulus with nanosecond pulsed electricfields (nsPEFs) in E4 squamous cell carcinoma, Apoptosis 16 (2011) 382–393.

[21] M. Stacey, J. Stickley, P. Fox, V. Statler, K. Schoenbach, S. Beebe, S. Buescher, Differentialeffects in cells exposed to ultra-short, high intensity electric fields: cell survival, DNAdamage, and cell cycle analysis, Mutat. Res./Genet. Toxicol. Environ. Mutagen. 542(2003) 65–75.

[22] S.J. Beebe, N.M. Sain, W. Ren, Induction of cell death mechanisms and apoptosis bynanosecond pulsed electric fields (nsPEFs), Cells 2 (2013) 136–162.

[23] M. Stacey, P. Fox, S. Buescher, J. Kolb, Nanosecond pulsed electric field inducedcytoskeleton, nuclear membrane and telomere damage adversely impact cellsurvival, Bioelectrochemistry 82 (2011) 131–134.

[24] X. Chen, J. Zhuang, J. Kolb, K. Schoenbach, S. Beebe, Long term survival of mice withhepatocellular carcinoma after pulse power ablationwith nanosecondpulsed electricfields, Technol. Cancer Res. Treat. 11 (2012) 83.

[25] G. Pucihar, J. Krmelj, M. Rebersek, T.B. Napotnik, D. Miklavcic, Equivalent pulseparameters for electroporation, IEEE Trans. Biomed. Eng. 58 (2011) 3279–3288.

[26] R. Nuccitelli, U. Pliquett, X. Chen,W. Ford, R. James Swanson, S.J. Beebe, J.F. Kolb, K.H.Schoenbach, Nanosecond pulsed electric fields cause melanomas to self-destruct,Biochem. Biophys. Res. Commun. 343 (2006) 351–360.

79M.B. Sano et al. / Bioelectrochemistry 100 (2014) 69–79

[27] W.H. Baldwin, B.W. Gregory, C.J. Osgood, K.H. Schoenbach, J.F. Kolb, Membranepermeability and cell survival after nanosecond pulsed-electric-field exposure-significance of exposure-media composition, IEEE Trans. Plasma Sci. 38 (2010)2948–2953.

[28] B.L. Ibey, A.G. Pakhomov, B.W. Gregory, V.A. Khorokhorina, C.C. Roth, M.A.Rassokhin, J.A. Bernhard, G.J. Wilmink, O.N. Pakhomova, Selective cytotoxicity ofintense nanosecond-duration electric pulses in mammalian cells, Biochim. Biophys.Acta-Gen. Subj. 1800 (2010) 1210–1219.

[29] K. Mitsutake, A. Satoh, S. Mine, K. Abe, S. Katsuki, H. Akiyama, Effect of pulsingsequence of nanosecond pulsed electric fields on viability of HeLa S3 Cells, IEEETrans. Dielectr. Electr. Insul. 19 (2012) 337–342.

[30] C.B. Arena, M.B. Sano, J.H. Rossmeisl Jr., J.L. Caldwell, P.A. Garcia, M.N. Rylander, R.V.Davalos,High-frequency irreversible electroporation (H-FIRE) for non-thermal ablationwithout muscle contraction, Biomed. Eng. Online 10 (2011).

[31] C.B. Arena, M.B. Sano, M.N. Rylander, R.V. Davalos, Theoretical considerations oftissue electroporation with high-frequency bipolar pulses, IEEE Trans. Biomed.Eng. 58 (2011) 1474–1482.

[32] J.C.Weaver, K.C. Smith, A.T. Esser, R.S. Son, T. Gowrishankar, A brief overviewof electro-poration pulse strength–duration space: a region where additional intracellular effectsare expected, Bioelectrochemistry (2012) 236–243.

[33] M. Bower, L. Sherwood, Y. Li, R. Martin, Irreversible electroporation of the pancreas:definitive local therapy without systemic effects, J. Surg. Oncol. 104 (2011) 22–28.

[34] G. Pucihar, T. Kotnik, B. Valič, D.Miklavčič, Numerical determination of transmembranevoltage induced on irregularly shaped cells, Ann. Biomed. Eng. 34 (2006) 642–652.

[35] T. Kotnik, D. Miklavčič, Theoretical evaluation of voltage inducement on internalmembranes of biological cells exposed to electric fields, Biophys. J. 90 (2006) 480–491.

[36] L.A. Flanagan, J. Lu, L. Wang, S.A. Marchenko, N.L. Jeon, A.P. Lee, E.S. Monuki, Uniquedielectric properties distinguish stem cells and their differentiated progeny, StemCells 26 (2008) 656–665.

[37] D. Clark, J. Greenwell, A. Harper, A.M. Sankey, T. Scratcherd, The electrical propertiesof resting and secreting pancreas, J. Physiol. 189 (1967) 247–260.

[38] J. von Burstin, S. Eser, M.C. Paul, B. Seidler, M. Brandl, M. Messer, A. von Werder,A. Schmidt, J. Mages, P. Pagel, E-cadherin regulates metastasis of pancreaticcancer in vivo and is suppressed by a SNAIL/HDAC1/HDAC2 repressor complex,Gastroenterology 137 (2009) 361.

[39] B. Seidler, A. Schmidt, U. Mayr, H. Nakhai, R.M. Schmid, G. Schneider, D. Saur, ACre-loxP-based mouse model for conditional somatic gene expression andknockdown in vivo by using avian retroviral vectors, Proc. Natl. Acad. Sci. 105(2008) 10137–10142.

[40] D. Saur, B. Seidler, G. Schneider, H. Algül, R. Beck, R. Senekowitsch–Schmidtke, M.Schwaiger, R.M. Schmid, CXCR4 expression increases liver and lung metastasis ina mouse model of pancreatic cancer, Gastroenterology 129 (2005) 1237–1250.

[41] M.J. PaszeK, N. Zahir, K.R. Johnson, J.N. Lakins, G.I. Rozenberg, A. Gefen, C.A. Reinhart-King, S.S. Margulies, M. Dembo, D. Boettiger, Tensional homeostasis and the malignantphenotype, Cancer Cell 8 (2005) 241–254.

[42] J. O'Brien, I. Wilson, T. Orton, F. Pognan, Investigation of the Alamar blue (resazurin)fluorescent dye for the assessment of mammalian cell cytotoxicity, Eur. J. Biochem.267 (2000) 5421–5426.

[43] K. Kinosita, I. Ashikawa, N. Saita, H. Yoshimura, H. Itoh, K. Nagayama, A. Ikegami,Electroporation of cell membrane visualized under a pulsed-laser fluorescencemicroscope, Biophys. J. 53 (1988) 1015–1019.

[44] K. Kinosita, T.Y. Tsong, Formation and Resealing of Pores of Controlled Sizes inHuman Erythrocyte Membrane, 1977.

[45] K. Kinosita, T.Y. Tsong, Voltage-induced pore formation and hemolysis of humanerythrocytes, Biochim. Biophy. Acta (BBA)-Biomembr. 471 (1977) 227–242.

[46] O.M. Nesin, O.N. Pakhomova, S. Xiao, A.G. Pakhomov, Manipulation of cell volume andmembranepore comparison following single cell permeabilizationwith60- and600-nselectric pulses, Biochim. Biophys. Acta Biomembr. 1808 (2011) 792–801.

[47] B. Alberts, D. Bray, J. Lewis, M. Raff, K. Roberts, J.D. Watson, Molecular Biology of theCell, 3rd edition Garland Science, New York, 1994.

[48] P.R. Gascoyne, R. Pethig, J.P. Burt, F.F. Becker, Membrane changes accompanyingthe induced differentiation of Friend murine erythroleukemia cells studied bydielectrophoresis, Biochim. Biophys. Acta (BBA)-Biomembr. 1149 (1993) 119–126.

[49] J. Yang, Y.Huang, X.J.Wang, X.B.Wang, F.F. Becker, P.R.C. Gascoyne,Dielectric propertiesof human leukocyte subpopulations determined by electrorotation as a cell separationcriterion, Biophys. J. 76 (1999) 3307–3314.

[50] I. Ermolina, Y. Polevaya, Y. Feldman, B.-Z. Ginzburg, M. Schlesinger, Study of normalandmalignant white blood cells by time domain dielectric spectroscopy, IEEE Trans.Dielectr. Electr. Insul. 8 (2001) 253–261.

[51] J. Gimsa, T. Müller, T. Schnelle, G. Fuhr, Dielectric spectroscopy of single humanerythrocytes at physiological ionic strength: dispersion of the cytoplasm, Biophys.J. 71 (1996) 495–506.

[52] K. Asami, Y. Takahashi, S. Takashima, Dielectric properties of mouse lymphocytesand erythrocytes, Biochim. Biophys. Acta (BBA)-Mol. Cell Res. 1010 (1989) 49–55.

[53] A.C. Sabuncu, J.A. Liu, S.J. Beebe, A. Beskok, Dielectrophoretic separation of mousemelanoma clones, Biomicrofluidics 4 (2010) 021101.

[54] F.H. Labeed, H.M. Coley, M.P. Hughes, Differences in the biophysical properties ofmembrane and cytoplasm of apoptotic cells revealed using dielectrophoresis,Biochim. Biophys. Acta (BBA)-Gen. Subj. 1760 (2006) 922–929.

[55] A. Ron, R.R. Singh, N. Fishelson, I. Shur, R. Socher, D. Benayahu, Y. Shacham-Diamand, Cell-based screening for membranal and cytoplasmatic markers usingdielectric spectroscopy, Biophys. Chem. 135 (2008) 59–68.

[56] H. Mulhall, F. Labeed, B. Kazmi, D. Costea, M. Hughes, M. Lewis, Cancer, pre-cancerand normal oral cells distinguished by dielectrophoresis, Anal. Bioanal. Chem. 401(2011) 2455–2463.

[57] J. Chen, Y. Zheng, Q. Tan, E. Shojaei-Baghini, Y.L. Zhang, J. Li, P. Prasad, L. You, X.Y.Wu, Y. Sun, Classification of cell types using a microfluidic device for mechanicaland electrical measurement on single cells, Lab Chip 11 (2011) 3174–3181.

[58] K. Helczynska, Å. Kronblad, A. Jögi, E. Nilsson, S. Beckman, G. Landberg, S. Påhlman,Hypoxia promotes a dedifferentiated phenotype in ductal breast carcinoma in situ,Cancer Res. 63 (2003) 1441–1444.

[59] A. Salmanzadeh, M.B. Sano, R.C. Gallo-Villanueva, P.C. Roberts, E.M. Schmelz, R.V.Davalos, Investigating dielectric properties of different stages of syngeneic murineovarian cancer cells, Biomicrofluidics 7 (2013) 011809.

[60] A. Maček-Lebar, D. Miklavčič, Cell electropermeabilization to small moleculesin vitro: control by pulse parameters, Radiol. Oncol. 35 (2001).

[61] K. Seibert, S.M. Shafie, T.J. Triche, J.J. Whang-Peng, S.J. O'Brien, J.H. Toney, K.K. Huff,M.E. Lippman, Clonal variation of MCF-7 breast cancer cells in vitro and in athymicnude mice, Cancer Res. 43 (1983) 2223–2239.

[62] Y. Shimizu, S. Kamoi, S. Amada, F. Akiyama, S.G. Silverberg, Toward the developmentof a universal grading system for ovarian epithelial carcinoma, Cancer 82 (1998)893–901.

[63] A. Malpica, M.T. Deavers, K. Lu, D.C. Bodurka, E.N. Atkinson, D.M. Gershenson, E.G.Silva, Grading ovarian serous carcinoma using a two-tier system, Am. J. Surg. Pathol.28 (2004) 496–504.

[64] S.G. Silverberg, Histopathologic grading of ovarian carcinoma: a reviewand proposal,Int. J. Gynecol. Pathol. 19 (2000) 7–15.

[65] A. Salmanzadeh, H. Kittur, M.B. Sano, P.C. Roberts, E.M. Schmelz, R.V. Davalos,Dielectrophoretic differentiation of mouse ovarian surface epithelial cells,macrophages, and fibroblasts using contactless dielectrophoresis, Biomicrofluidics6 (2012) 024104.

[66] A. Salmanzadeh, E.S. Elvington, P.C. Roberts, E.M. Schmelz, R.V. Davalos, sphingolipidmetabolites modulate dielectric characteristics of cells in a mouse ovarian cancerprogression model, Integr. Biol. 5 (6) (2013) 843–852.

[67] J. Rubinsky, G. Onik, P. Mikus, B. Rubinsky, Optimal parameters for the destruction ofprostate cancer using irreversible electroporation, J. Urol. 180 (2008) 2668–2674.

[68] G. Onik, B. Rubinsky, Irreversible electroporation: first patient experience focaltherapy of prostate cancer, in: B. Rubinsky (Ed.), Irreversible Electroporation,Springer, Berlin Heidelberg, 2010, pp. 235–247.

[69] R. Cannon, S. Ellis, D. Hayes, G. Narayanan, R.C. Martin, Safety and early efficacy ofirreversible electroporation for hepatic tumors in proximity to vital structures, J.Surg. Oncol. 107 (5) (2013) 544–549 (April).

[70] M.B. Sano, E.A. Henslee, E. Schmelz, R.V. Davalos, Contactless dielectrophoreticspectroscopy: examination of the dielectric properties of cells found in blood,Electrophoresis 32 (2011) 3164–3171.