Embed Size (px)
Citation preview
Biomed. Phys. Eng. Express 3 (2017) 015005 doi:10.1088/2057-1976/3/1/015005
PAPER
Dielectrophoresis as a single cell characterization method forbacteria*
MElitas1,4,5, NDhar1, K Schneider1, AValero2,3, T Braschler2, J DMcKinney1 and PRenaud2
1 School of Life Sciences, École polytechnique fédérale de Lausanne (EFPL), 1015 Lausanne, Switzerland2 Laboratory ofMicrosystems, École polytechnique fédérale de Lausanne (EFPL), 1015 Lausanne, Switzerland3 CICnanoGUNE, TolosaHiribidea, 76 San Sebastian, Spain4 Present address:Mechatronics Program, Sabanci University (SU), Istanbul, Turkey.5 Author towhomany correspondence should be addressed.
E-mail:[email protected], [email protected], [email protected], [email protected], [email protected], [email protected] and [email protected]
Keywords: dielectrophoresis, antibiotic, single-cell, characterization,modeling, bacteria
AbstractReal-time, quantitative characterization of cells at single-cell resolution, particularly whilemaintain-ing their intrinsic properties andwithout affecting cellular processes, is of primary importance inmodern biological assays. Dielectrophoresis is a label-free, real-time, and quantitative technique, andis amenable to integrationwith other techniques, thus providing new and powerful tools for biologyandmedicine. In this studywe present dielectrophoresis as a characterization tool forMycobacteriumsmegmatis single cells. Understanding howphenotypically variantM. smegmatis cells responddielectrophoretically when subject to the same electricfield, could reveal underlyingmembranealteringmechanisms related to cell death, drug-tolerance, and drug-resistance. In this study, wedielectrophoretically characterized live, heat-treated and antibiotic-treated bacteria. Our resultspresent quantifications of cellular behaviors associatedwithmembrane-specific cell damages anddemonstrate adequacy of dielectrophoretic devices in point-of-care diagnostic andmonitoring forbacterial infections.
Introduction
Since its introduction in the 1950s and first applicationto bacteria in the 1970s, dielectrophoresis (DEP) hasbecome an active area of research as a method fordetection, concentration, separation, identification,and manipulation of cells and molecules [1–6]. DEPcan be used to measure the dielectric properties ofcells, providing valuable insight into cellular physiol-ogy without causing significant cellular damage ordeath. Thus, DEP devices have been used to character-ize and separate mammalian cells, microorganisms[7], plant cells [8], viruses [9], DNA [10], RNA [11],proteins [12–14], as well as to distinguish specificmammalian cells for drug screening, isolate infectedcells from blood [15], and identify membrane proper-ties of particles and cells [16, 17]. DEP can also be used
to make real-time measurements of cellular para-meters in order to distinguish live cells from damagedor dead ones [18]. Finally, it can also be applied to drugand phenotypic screening [19, 22].
Analysis of bacteria using lab-on-a-chip devices ischallenging due to their small cell size, cell-cell adher-ence, biofilm formation, and biosafety considerations.However, at the same time, DEP has emerged as apowerful technique for microbiology research, usedalone or in combination with traditional approachessuch as antibiotic sensitivity testing [21, 22], DNAmanipulation [23]. Selective detection of viable bac-teria is now possible through DEP impedance mea-surements via capture on an electrode array [24],which can even be coupled with antigen-antibodyinteractions [25]. DEP has also been used to distin-guish wild-type bacterial cells frommutant derivativespossessing altered dielectric properties [21, 26], and itseffects on growth, viability, and immuno-reactivity ofbacteria have been previously studied [27].
RECEIVED
6 July 2016
REVISED
6October 2016
ACCEPTED FOR PUBLICATION
7November 2016
PUBLISHED
6 January 2017
*Preliminary results of the current paper were presented at Micro
TAS 2009, Jeju, SouthKorea.
© 2017 IOPPublishing Ltd
Recently, the DEP properties of bacteria have beeninvestigated extensively, especially in the contexts ofdrug assessment [28], antibiotic susceptibility [29] anddormancy [30]. Specifically, subpopulations of cellssuch as persisters, dormant, and non-culturable cellsare responsible for the drug recalcitrance of manyinfectious diseases. To better understand the mechan-isms of drug-cell interactions that is undermining pro-gress in this area, we carried out dielectrophoreticanalysis of the non-pathogenic mycobacteria, M.smegmatis. Dielectrophoretic characterization of wild-type (WT) M. smegmatis and ethambutol-treated M.smegmatis has already been reported, where it wasshown that ethambutol-treated bacteria exhibit a di-electric response at higher pDEP frequencies thanwild-type cells [20]. Dielectrophoretic approaches forinvestigating dormant mycobacterial cells have con-centrated on separation of dormant (stationary phase,i.e., nongrowing) and actively growing M. smegmatiscells [30], as well as monitoring the resuscitation ofdormant M. smegmatis in the DEP-generated micro-bial aggregates [30].
However, more recently, DEP methods are beingexplored as purification tools for separating sub-populations of antibiotic-treated bacterial cultures,allowing characterization of surviving cells and anti-biotic-killing mechanisms [31]. Here, we present thedielectrophoretic behavior of live, heat-killed, andantibiotic-treated M. smegmatis cells at single-cellresolution. We begin with a brief introduction to die-lectrophoretic forces, and follow with a simple single-shell sphericalmodel forM. smegmatis bacterium, use-ful for predicting the dielectrophoretic behavior ofboth live and dead cells. Next, we present the results ofdielectrophoretic characterization experiments forlive and heat-killed cells and the dielectrophoreticresponses of Isoniazid (INH)-treated cells. Finally, wecompare the results for live, INH-treated and heat-kil-ledM. smegmatis cells.
Dielectrophoresis
DEP forces ( FDEP) on a particle depend on themagnitude and nonuniformity of an externally appliedelectric field, as well as the physical and electricalparameters of the surroundingmedium and of the cell,such as conductivity and permeability, according toequation (1),
pe= [ ] ( )F r f E2 Re , 1DEP m3
CM rms2
where r is the radius of the cell, Erms2 is the root mean
square of the electric field, em is the permittivity of themedium, and [ ]fRe CM is the real part of the Clausius–Mossotti factor ( )f ;CM this defines the effective polar-izability of the cell as
=e e
e e
-
+[ ] ( )( )
( )
* *
* *fRe , 2CM 2
p m
p m
where e*p is the complex permittivity of the particle,
and e*m that of the media. Finally, the complexpermittivity e* is given as
e e= - sp
( )* 3j
f2
and thus depends on the permittivity e( ) and con-ductivity s( ) of the cell or themedia and the frequencyf of the applied electric field. j represents the imaginarynumber -1 . [ ]fRe CM ranges from−0.5 to 1, wherepositive values denote cells migrating towards regionsof high field strength (positive dielectrophoresis,pDEP), while negative values of [ ]fRe CM denote theopposite behavior, with cells moving toward regionsof low or no field strength (negative dielectrophoresis,nDEP). Whether a cell displays nDEP or pDEPdepends on polarizability of the cell with respect to thesurrounding medium. The frequency at which thedielectrophoretic force is zero =( [ ] )fRe 0CM isdenoted as the crossover frequency.
Modeling
In order to characterize DEP conditions, DEP bufferconductivity, electric field strength, and field fre-quency were analyzed both experimentally and using atheoretical single-shellmodel.
A theoretical single-shell spheroidalmodel forM.smegmatis cellsM. smegmatis is a rod-shaped bacterium with acomplex cell wall that includes both an inner and anouter membrane, as shown in figure 1 [31, 32]. Thus, aprolate ellipsoidal dielectric model would provide thebest fit for in silico analysis of the dielectric propertiesof this bacterium. Nonetheless, for reasons of simpli-city and comprehensibility, a single-shell spheroidal[33, 34]model was chosen. In this simplified model, acell is considered to be a sphere of conductingcytoplasm surrounded by an insulating membrane.The cell membrane and cell wall are represented by asingle term and modeled according to their resistance(R ,cell wall equation (4)), their capacitance (C ,cell wall
equation (5)), and their impedance (Z ,cell wall
equations (6)–(7)). The cytoplasm was modeled in thesameway and is represented in equations (8)–(11).
=s( )( ) ( )R , 4k d
rcell wallcell wall
= e( )( ) ( )C , 5k
r
dcell wallcell wall
=+
w
w
( )( )Z
R, 6
R
j C
j C
cell wallcell wall
1
cell wall
cell wall
cell wall
=we s+
( )Z , 7kd
jcell wallcell wall cell wall
=s( ) ( )R , 8k
cytoplasmcytoplasm
=e( ) ( )C , 9
kcytoplasmcytoplasm
2
Biomed. Phys. Eng. Express 3 (2017) 015005 MElitas et al
=+
w
w
( )( )Z
R, 10
R
j C
j C
cytoplasmcytoplasm
1
cytoplasm
cytoplasm
cytoplasm
=we s+
( )Z . 11k
jcytoplasmcytoplasm cytoplasm
Thefinal cellmodel is comprised the serial impedancesof Zcell wall and Z ,cytoplasm and is denoted ZCell:
= + ( )Z Z Z , 12Cell cell wall cytoplasm
=
+
we s
we s
+
+( )
( )
( )
Z
, 13
d
r
k
j
k
j
Cellcell wall cell wall
cytoplasm cytoplasm
e =w
( )* , 14k
j Zcell
where k is a cell constant, d is the cell wall thickness, ris the cell radius, R is the resistance, C is thecapacitance, s is the conductivity, and e is thepermittivity [32, 33]. The dielectric parameters appliedforM. smegmatis are listed in table 1 [26, 33, 35]. Thereal part of the Clausius–Mossotti factor ( [ ])fRe CMwas calculated inMATLAB using the parameters fromtable 1. The results obtained are shown in figure 2. Insummary, while themodeled live cells exhibited pDEPpast the crossover frequency of 1 MHz, the modeleddead cells exhibited nDEP in the entire range of0.1–10MHz.
Materials andmethods
Buffer preparationTo prepare the gradient of conductivities (rangingfrom 2 to 500 μS m−1, measured by a conductivitymeter; Cole–Parmer Instruments), increasingamounts of phosphate buffered saline (PBS, Gibco)were mixed in double distilled water. The dispersalagent Tween-80 (Sigma-Aldrich), was added at 0.05%to avoid formation of cell clumps.
Cell growth and preparationBatch cultures of wild-type M. smegmatis strainmc2155 were grown overnight in standard 7H9 media(BD/Difco) with shaking (200 rpm), to an opticaldensity (OD) of 0.5, measured as absorbance at600 nm (Thermo scientific, Biomates). To distinguishlive cells from dead cells, M. smegmatis recombinantstrains expressingGFP or dsRed2 fluorescent proteins,specifically from a strong constitutive promoter wereused [36]. Before introducing the cells into DEPdevice, both live and dead cells were collected bycentrifugation (≈900×g, 5 min), washed three timesin DEP buffers, and finally resuspended in therespective DEP buffer. Dead cells were stained withpropidium iodide (ThermoFisher Scientific) at a finalconcentration of 1 μg ml−1. The cell suspension wasfiltered using a 0.5 μm filter (Millex-SV) to remove cellagglomerates.
Preparation of heat-killed cellsCells were taken from exponentially growing cultureand incubated at 80 °C for 1 h to obtain heat-killedcells. Theywere then collected and prepared as above.
Preparation of antibiotic-treated cellsBatch cultures of GFP-expressing M. smegmatis cells[36] were grown in standard 7H9 media (BD/Difco)with shaking (200 rpm), to an optical density of 0.05(A600nm), then exposed to isoniazid (INH)(50 μg ml−1) for 48 h at 37 °Cwith shaking (200 rpm).
Figure 1. Single-shell dielectricmodel andmembrane structure ofM. smegmatis. (a)M. smegmatis, recombinant strains expressingGFP or dsRed2 fluorescent proteins, grow in basalmedium (7H9). (b)Themycobacterial cell envelope structure, specificallyMycolicacids (1), Arabinogalactan (2), Peptidoglycan (3), and Bylipid layer (4). (c)The simplified single shell dielectricmodel, whereZcell wall isthe impedance of the cellmembrane andwall andZcytoplasm is the impedance of the cytoplasm. The green line represents the cross-section region.
Table 1.The dielectric parameters used formodeling andsimulation of [ ]fRe CM factors.
Parameters Live cell Dead cell
scytoplasm-0.48 S m 1 s´ + -0.8 0.01 S mmedium
1
scell wall- -10 S m6 1 - -10 S m6 1
ecytoplasm e´50 0 e´50 0
ecell wall e´10 0 e´10 0
e0 ´ -8.85 10 12 ´ -8.85 10 12
d ´ -8 10 m9 ´ -8 10 m9
r ´ -500 10 m9 ´ -500 10 m9
k -r 1 -r 1
emedium e´80 0 e´80 0
3
Biomed. Phys. Eng. Express 3 (2017) 015005 MElitas et al
Before use, cells were collected and prepared asdescribed above.
DEPdevice preparationDielectrophoretic forces were generated using 2D-metal electrodes on a glass substrate as reportedpreviously [3, 34, 37]. An array of 2×15 platinumelectrodes and structured SU-8 line the main channel(width: 20 μm, height: 20 μm) that cells flow throughand where dielectrophoretic forces are applied. Thecells were characterized according to deviation fromthe center of the main channel as shown in figure 3.The frequencies of the applied AC field varied from100 to 15MHz and the AC field potentials varied from0.1 Vrms to 10.5 V .rms Before introducing the cells intothe device, the device was rinsed with low-conductivebuffer that was used for the cell characterization. Thedevice was stringently cleaned using detergent anddouble distilledwater after each experiment.
Data acquisitionCells were monitored using an Oylmpus IX81S1F-ZDC motorized inverted fluorescence microscopeequipped with 100× objective and a HamamatsuORCA/ER CCD camera. All acquired images wereintegrated intomovies and analyzed using ImageJ.
Experimental procedureAfter cell preparation, 25 μl of the cell suspension wasintroduced into the device. First, the cells were focusedat the center of the separation channel by applying lowAC frequencies from both sides of the electrode array.Then, a higher AC frequency was superimposed fromone side of the electrode array. Cells, based on theirintrinsic properties, were positioned at an equilibriumpoint, either migrating towards regions of high fieldstrength (pDEP) or migrating away from them(nDEP). The equilibrium position was used as readout
to assess the dielectric properties of the cells, specific tocell type. The crossover frequencies for live and deadcells were defined as the frequencies where DEP forcesacting on a cell switch from negative to positive (forlive cells) or vice-versa (for dead cells). The experi-ments were repeated with varyingmedium conductiv-ities and a range of frequencies to determine thecombination of optimum conductivity and frequencyvalues for characterization of subpopulations of cells.
Results
Experimentswith live andheat-treated cellsDielectrophoretic characterization of live and heat-killed cells were performed separately as illustrated infigure 3. In a medium with conductivity of 2.7 μS m−1
and with a frequency range above 1 MHz, the live andheat-treated dead cells exhibited opposite DEP beha-viors. Live cells, were pulled towards the high field,exhibiting pDEP, and dead cells were pulled towardthe low field, signaling nDEP, indicating the potentialfor dielectrophoretic characterization of mixed liveand dead populations of M. smegmatis at single-cellresolution.
A representative set of images depicting the beha-vior of GFP-expressing live cells in the presence ofhigh frequency signals is shown in figure 3. This cellexhibited nDEP in the lower frequencies (the x-axis:applied AC frequency), and was pulled −8 μm alongthe channel. When the frequency was increased, thecell instead exhibited pDEP and was pulled +9 μmalong the channel. Here, the dead cells were not pre-sented due to the fact that the Clausuis Mossotti factorwas always negative for dead cells. Thus, the dead cellswere experiencing nDEP forces all the time and theywere repelled by the electrodes for the whole fre-quency range.
Figure 2. Simulation of theClausuis–Mossotti factors for both live (red) and dead (blue) cells in 2.7 μS m−1DEP buffer and 0.1 –10 MHz frequency range. Live cells have a crossover frequency around 1 MHz,whereas dead cells never cross the pDEP region.
4
Biomed. Phys. Eng. Express 3 (2017) 015005 MElitas et al
Experimentswith isoniazid-treated cellsThe dielectric response of the INH-treated cells to anon-uniform electric field was also explored. Whilethe crossover frequency of untreated M. smegmatiscells was consistently around 500 kHz, INH-treated M. smegmatis cells exhibited pDEP in higherfrequencies than untreated cells, owing to its
higher crossover frequency, PI-stained (dead) cellsexhibited nDEP at low frequencies, but when thefrequency increased they did not show clear pDEP,and were likely not affected by DEP force in410 μS cm−1 buffer in the scanned frequency range(0.1–15MHz). Figure 4 summarizes the results ofthese experiments.
Figure 3.Dielectrophoretic characterization ofM. smegmatis using crossover frequencies. Single cell suspension ofM. smegmatis cellswas introduced in a low conductive buffer (2.7 μS m−1). Two low frequency signals were used to center the cells along themidline ofthe 20 μmwide channel (dashed-yellow line). Then, a high frequency signal ranging from0.1 to 15 MHzwas applied from left side ofthe channel. (a)The translationalmovement (μm) of the live cells relative to themidline of the channel is plotted versus frequencies(MHz). (b)Using the obtained displacement values, theClausuisMossotti factors for live cells are presented. (c)Movement of aGFP-expressing live cell (white) in themicrofluidic channel under the nonuniform electricfield; the dashed-green lines are the borders ofthemicrofluidic channel.
Figure 4.TheDEP behaviors of wild-type and INH-treatedM. smegmatis cells. Untreated cells exhibited crossover around 500 kHz,whereas INH-treated cells did so around 900 kHz in 410 μS cm−1 buffer from100 kHz to 15 MHz.
5
Biomed. Phys. Eng. Express 3 (2017) 015005 MElitas et al
Conclusion
DEP can be used to measure the dielectric propertiesof cells, which provides important insights intocellular physiology while avoiding significant cellulardamage or death. Although it is an old and veryestablished method, it becomes more and morepowerful and adequate with advancements in technol-ogy. In order to obtain direct, accurate, real-time, andquantitativemeasurements related to intrinsic proper-ties of cells in a label-free way, it presents adequatemethods to complete traditional bulk assays in biol-ogy. The aim of this work was to use DEP tocharacterize phenotypically variant bacterial subpopu-lations based on their intrinsic dielectric propertiesunder antibiotic exposure and heat treatment. Wild-type M. smegmatis cells do not have any dielectro-phoretic difference prior to heat- or drug-treatment inour experiments. Our efforts were focused on optimiz-ing DEP characterization in M. smegmatis cells basedon only small changes in their intrinsic properties,rather than differences in their size, shape, or volume.The recent dielectrophoretic monitoring and separa-tion technique fromSu and his co-workers implemen-ted dielectrophoteric method on C. difficile strainswith differing surface-layer properties. Therefore,drug treatment enhanced DEP-based separation ofmorphologically different cells [21]. The dielectro-phoretic assay of bacterial resistance to antibioticsdeveloped by Johari et al also performed experimentsusing S. epidermidis strains with different cell wallpermittivity that significantly contributed collection ofantibiotic-sensitive and antibiotic-resistant cells [22].
Furthermore, we presented very simple and com-prehensive single-shell spheroidal model in this work.Althoughmore complexmulti-shellmodels are repor-ted in the literature [20], using this simple model weaimed to predict the dielectrophoretic response of liveand damaged cells in the frequency range of 100 kHz–1.5 MHz. Our experimental results were consistentwith the obtained data from a single-shell spheroidalmodel. When the models becomes complex, theyrequire high computing performance and long time toperform calculations, therefore, comparing resultswith real-time data to get insights for the predicationof experimentsmight become cumbersome.
When wild-type, INH-treated, and heat-treatedM. smegmatis cells were dielectropherically character-ized, the heat-killed cells do not experience strongDEP forces compared to wild-type cells due to theirimpaired cell membrane composition, they are eitherat nDEP, or at crossover frequency regimes where theDEP forces are vanishingly small. On the other hand,the crossover frequency values for antibiotic-treatedcells increases compared to wild-type cells, they showpDEP behavior at higher frequencies. INH targetsmycolic acids biosynthesis in mycobacteria, therefore,impaired permeability of the cell membrane decreasesthe complex permeability of the cells, as reported for
the dielectrophoretic characterization of ethambutoltreated mycobacteria [20], and pDEP responses of thecells are observed at higher frequencies.
Moreover, our study demonstrates that DEP plat-forms can be used to monitor the changes in dielec-trophoretic behavior of bacteria as a single-cell orpopulation level over time via time-lapse microscopy.This technique can be used to address a variety of out-standing questions in cell behavior, such as cell culturesynchronization based on cell-cycle phase or cell age,diagnosis of infected cells (macrophages), cell-to-cellinteractions between different cell types, and inter-cellular communication. Last but not least, DEP forcescan also be exploited to immobilize motile cells with-out inducing any mechanical stress, by allowingmicroscopic observation of their behavior without theneed for complexmicrofluidic platforms.
Acknowledgments
We thank Kristin Irwin, PhD, for her assistance in thepreparation and editing of the manuscript. We grate-fully acknowledge core support from the EPFL, SU,and research grants from the United States NationalInstitutes of Health and the Bill & Melinda GatesFoundation.
References
[1] PohlHA1978Dielectrophoresis: the behavior of neutral matterin nonuniform electric field (NewYork: CambridgeUniversityPress)
[2] Lapizco-Encinas BH, Simmons BA, Cummings EB andHenkoY 2004 Insulator-based dielectrophoresis for theselective concentration and separation of live bacteria inwaterElectrophoresis 25 1695–704
[3] DemierreN, Braschler T,Muller R andRenaud P 2007Focusing and continuous separation of cells in amicrofluidicdevice using lateral dielectrophoresis Solid-State Sensors,Actuators andMicrosystems Conf. and Transducers pp 1777–80
[4] Yan S, Zhang J, PanC, YuanD, Alici G,DuH, ZhuY and LiW2015An integrated dielectrophoreis-active hydrophoreticmicrohip for continuous particle filtration and separationJ.Micromech.Microeng. 25 084010–9
[5] Chi L andMorganH2000Design and fabrication of travellingwave dielectrophoresis structues J.Micromech.Microeng. 1072–9
[6] ChurchC, Zhu J, Nieto J, KetenG, Ilbarra E andXuanX2010Continuous particle separation in a serpentinemicrochannelvia negative and positive dielectrophoretic focusingJ.Micromech.Microeng. 20 065011–7
[7] Peitz I and van LeeuwenR 2010 Single-cell bacteria growthmonitoring by automatedDEP-facilitated image analysis LabChip 10 2944–51
[8] FalokunCD,Mavituna F andMarkzGH2003ACelektrokinetic characterisation and separation of cells withhigh and low embriyogenic potential in suspension culturesof carrot (Daucus carota)Plant Cell, TissueOrgan Cult. 75261–72
[9] HughesMP,MorganHandRixon F J 2002Measuring thedielectric properties of herpes simplex virus type 1 virions withdielectrophoresisBiochimica et Biophys. Acta (BBA)—Gen.Subjects 1571 1–8
[10] ChouCF, Tegenfeldt JO, BakajinO, Chan S S, Cox EC,DarntonN,Duke T andAusitn RH2002 Electrodeless
6
Biomed. Phys. Eng. Express 3 (2017) 015005 MElitas et al
dielectrophoresis of single- and double-strandedDNABiophys. J. 83 2170–9
[11] GiraudG et al 2011Dielectrophoreticmanipulation ofribosomal RNABiomicrofluidics 5 24116
[12] Clarke RW,White S S, ZhouD, Ying L andKlenermanD2005Trapping of proteins under physiological conditions in ananopipetteAngew. Chem. 117 3813–6
[13] Clarke RWAP, JoeD, Ying L andKlenermanD2007 Surfaceconductivity of biologicalmacromoleculesmeasured bynanopipette dielectrophoresis Phys. Rev. Lett. 98 198102
[14] Lapizco-Encinas BH,Ozuna-Chacón S andRito-PalomaresM2008 Proteinmanipulationwith insulator-baseddielectrophoresis and direct current electricfieldsJ. Chromatogr.A 1206 45–51
[15] Gascoyne P,Mahidol C, RuchirawatM, Satayavivad J,Watcharasit P andBecker F F 2002Microsample preparationby dielectrophoresis: isolation ofmalaria LabChip 2 70–5
[16] Vykoukal J, Vykoukal DM, Sharma S, Becker F F andGascoyne PRC2003Dielectrically addressablemicrospheresengineered using self-assembledmonolayers Langmuir 192425–33
[17] Oblak J, KrizajD, Amon S,Macek-Lebar A andMiklavcicD2007 Feasibility study for cell electroporation detectin andseparation bymeans of dielectrophoresisBioelectrochemistry71 164–71
[18] Patel P andMarkxGH2008Dielectricmeasurement of celldeathEnzymeMicrobial Technol. 43 463–70
[19] Hsiung L-C et al 2011Dielectrophoresis-based cellularmicroarray chip for anticancer drug screening in perfusionmicroenvironments LabChip 11 2333–42
[20] Hawkins BG,HuangC, Arasanipalai S andKirby B J 2011Automated dielectrophoretic characterization ofmycobacterium smegmatisAnal. Chem. 83 3507–15
[21] SuY-H,WarrenCA,Guerrant R L and SwamiNS 2014Dielectrophoreticmonitoring and interstrain separatin ofClostridium difficile based on their s(surface)-layers intactAnal.Chem. 86 10855–20863
[22] Johari J, Hubner Y,Hill J C, Dale JWandHughesMP2003Dielectrophoretic assay of bacterial resistance to antibioticsPhys.Med. Biol. 48 193–8
[23] Regtmeier J, DuongTT, EichhornR, Anselmetti D andRosA2007Dielectrophoreticmanipulation ofDNA: separation andpolarizabilityAnal. Chem. 79 3925–32
[24] Suehiro J,Hamada R,NoutomiD, ShutouMandHaraM2003Selective detection of viable bacteria using dielectrophoreticimpedancemeasurementmethod J. Electrost. 57 157–68
[25] Suehiro J, Yatsunami R,Hamada R andHaraM1999Quantitative estimation of biological cell concentration
suspended in aqueousmediumby using dielectrophoreticimpedancemeasurementmethod J. Phys. D: Appl. Phys. 322814–20
[26] CastellarnauM, Errachid A,MadridC, Juarez A and Samitier J2006Dielectrophoresis as a tool to characterize anddifferentiate isogenicmutants ofEscherichia coli Biophys. J. 913937–45
[27] Yang L, Banada P, Bhunia A andBashir R 2008 Effects ofdielectrophoresis on growth, viability and immuno-reactivityof Listeriamonocytogenes J. Biol. Eng. 2 6
[28] Hoettges K F,Hubner Y, Broche LM,Ogin S L, KassGENandHughesMP2008Dielectrophoresis-activatedmultiwell platefor label-free high-throughput drug assessmentAnal. Chem.80 2063–8
[29] ChungCC,Cheng I F, ChenH-M,KanH-C, YangW-H andChangH-C 2012 Screening of antibiotic susceptibility toβ-lactam-induced elongation of gram-negative bacteria based ondielectrophoresisAnal. Chem. 84 3347–54
[30] ZhuK, Kaprelyants A S, Salina EG, SchulerM andMarkxGH2010Construction by dielectrophoresis ofmicrobialaggregates for the study of bacterial cell dormancyBiomicrofluidics 4 022810
[31] ElitasM,Martinez-Duarte R, DharN,McKinney JD andRenaudP 2014Dielectrophoresis-based purification ofantibiotic-treated bacterial subpopulations LabChip 14 1850
[32] Zuber B, ChamiM,HoussinC,Dubochet J, GriffithsG andDafféM2008Direct visualization of the outermembrane ofmycobacteria and corynebacteria in their native stateJ. Bacteriol. 190 5672–80
[33] HoffmannC, Leis A,NiederweisM, Plitzko JM andEngelhardtH 2008Disclosure of themycobacterial outermembrane: cryo-electron tomography and vitreous sectionsreveal the lipid bilayer structureProc. Natl Acad. Sci. 1053963–7
[34] Braschler T 2009Controlled entrapment of cells in hydrogelson chip and cell sorting by dielectrophoresis using liquidelectrodes Doctoral ThesisÉcole polytechnique fédérale deLausanne (EFPL)
[35] ElitasM2012 Single-cell analysis ofmycobacterial persistenceusingmicrofabricated toolsDoctoral Thesis Écolepolytechnique fédérale de Lausanne (EFPL)
[36] Wakamato Y,DharN, Chait R, Schneider K, Signorino-Gelo F,Leibler S andMcKinney JD 2013Dynamic peristence ofantibiotic-stressedmycobacteria Science 339 91–5
[37] Braschler T,DemierreN,Nascimento E, Silva T,Oliva AG andRenaudP 2007Continuous separation of cells by balanceddielectrophoretic forces atmultiple frequencies LabChip 8280–6
7
Biomed. Phys. Eng. Express 3 (2017) 015005 MElitas et al
