the changing dielectric properties of cho cells can be...

The Changing Dielectric Properties of CHO CellsCan Be Used to Determine Early Apoptotic Eventsin a Bioprocess

Katrin Braasch,1 Marija Nikolic-Jaric,2 Tim Cabel,2 Elham Salimi,2 Greg E. Bridges,2

Doug J. Thomson,2 Michael Butler1

1Department of Microbiology, University of Manitoba, Winnipeg, Manitoba, R3T 2N2,

Canada; telephone: þ204-474-6543; fax: þ204-474-7603; e-mail: [email protected] of Electrical and Computer Engineering, University of Manitoba, Winnipeg,

Manitoba, Canada

ABSTRACT: To ensure maximum productivity of recombi-nant proteins it is desirable to prolong cell viability during amammalian cell bioprocess, and therefore important tocarefully monitor cell density and viability. In this study, fivedifferent and independent methods of monitoring wereapplied to Chinese hamster ovary (CHO) cells grown in abatch culture in a controlled bioreactor to determine celldensity and/or cell viability. They included: a particlecounter, trypan blue exclusion (Cedex), an in situ bulkcapacitance probe, an off-line fluorescent flow cytometer,and a prototype dielectrophoretic (DEP) cytometer. Thesevarious techniques gave similar values during the exponen-tial growth phase. However, beyond the exponential growthphase the viability measurements diverged. Fluorescent flowcytometry with a range of fluorescent markers was used toinvestigate this divergence and to establish the progress ofcell apoptosis: the cell density estimates by the intermediatestage apoptosis assay agreed with those obtained by the bulkcapacitance probe and the early stage apoptosis assayviability measurements correlated well with the DEP cytom-eter. The trypan blue assay showed higher estimates of viablecell density and viability compared to the capacitance probeor the DEP cytometer. The DEP cytometer measures thedielectric properties of individual cells and identified at leasttwo populations of cells, each with a distinct polarizability.As verified by comparison with the Nexin assay, one popu-lation was associated with viable (non-apoptotic) cells andthe other with apoptotic cells. From the end of the expo-nential through the stationary and decline stages there was agradual shift of cell count from the viable into the apoptoticpopulation. However, the two populations maintained theirindividual dielectric properties throughout this shift. This

leads to the conclusion that changes in bulk dielectricproperties of cultures might be better modeled as shifts incells between different dielectric sub-populations, ratherthan assuming a homogeneous dielectric population. Thisshows that bulk dielectric probes are sensitive to the earlyapoptotic changes in cells. DEP cytometry offers a novel andunique technology for analyzing and characterizing mam-malian cells based on their dielectric properties, and suggestsa potential application of the device as a low-cost, label-free,electronic monitor of physiological changes in cells.

Biotechnol. Bioeng. 2013;110: 2902–2914.

� 2013 Wiley Periodicals, Inc.

KEYWORDS: dielectrophoresis; capacitance; apoptosis;CHO; batch culture; flow cytometry

Introduction

Within the last decade there has been a rapid and increasingdemand for the large-scale production of glycoproteins frommammalian cell culture processes (Butler, 2005). Thisdemand is driven by the application of these molecules asbiopharmaceuticals for unmet medical needs (Walsh, 2010).To meet this high demand significant improvements havebeen made in cell line engineering including mammalianexpression vectors (Bebbington et al., 1992; Lucaset al., 1996) and the addition of anti-apoptotic genes(Kim and Lee, 2002; Mastrangelo et al., 2000; Teyet al., 2000) to enhance specific and volumetric productivity.In media development the ability to formulate media basedon the cells metabolic needs and the addition of sodiumbutyrate (Mimura et al., 2001) as well as other smallchemicals has shown to increase the recombinant proteinproduction in a bioprocess. Most importantly thesedevelopments in bioprocess techniques have improvedtiters. By changing from batch to optimized fed-batchprocesses cells stay metabolically active over extendedperiods of culture time increasing volumetric productivityup to 1–5 g/L for antibodies (De Jesus and Wurm, 2011; Lim

Braasch and Nikolic-Jaric contributed equally to this work.

Correspondence to: M. Butler

Contract grant sponsor: NSERC

Contract grant sponsor: Canada Foundation for Innovation (CFI)

Contract grant sponsor: Western Economic Diversification Canada (WD)

Contract grant sponsor: Canadian Microelectronics Corporation (CMC) Microsystems

Received 1 November 2012; Revision received 23 May 2013; Accepted 3 June 2013

Accepted manuscript online 1 July 2013;

Article first published online 3 July 2013 in Wiley Online Library

(http://onlinelibrary.wiley.com/doi/10.1002/bit.24976/abstract).

DOI 10.1002/bit.24976

ARTICLE

2902 Biotechnology and Bioengineering, Vol. 110, No. 11, November, 2013 � 2013 Wiley Periodicals, Inc.

et al., 2010). While these improvements increase cell specificand volumetric productivity the monitoring of cell viabilityduring a bioprocess is still pivotal to determining the endpoint of a bioprocess.Two commonly used methods to monitor the cell density

and the viability in a bioprocess are electronic particlecounting and image analysis with the use of specific dyes. TheCoulter counter registers electronically the presence ofparticles at a predetermined size and is a rapid methodthat can give the total cell density of a cell suspension. Thetrypan blue exclusion method, in combination with ahemocytometer or an automated image analyzer, allows atotal cell count as well as differentiation between “viable” and“non-viable” cells.Themeasurement of viable cells is important because these

are the metabolically active producer cells. However, theconcept of “viability,” is complex and depends critically onthe cell characteristics used to define it (Browne and Al-Rubeai, 2011). Commonly used assays, such as the trypanblue exclusionmethod, associate loss of viability with a loss ofcell membrane integrity as measured by the uptake of highmolecular weight dyes. However, specific metabolic changescan be identified that precede this and can be predictive of thereduced ability of cells to grow and divide when they entersenescence.An alternative technique for measuring changes in viable

cell density over the course of a bioprocess is by electricalcapacitance (Carvell and Dowd, 2006). A sterilizablecapacitance probe commercialized as a “biomass monitor”has been available over the last decade. Bulk capacitance andconductance of yeast and mammalian cell cultures can bemonitored by a dielectric probe positioned within thebioreactor (Ansorge et al., 2010; Ducommun et al., 2002;Knabben et al., 2010; Opel et al., 2010; Tibayrenc et al., 2011).The monitored capacitance of the bulk cell suspension, C, is afunction of the effective dielectric permittivity of thesuspension, e which is in turn a function of the biovolume.The dominant response of the cell to an electrical field ischarge polarization which diminishes as the frequency of thefield increases between 0.1 and 20MHz, resulting in a largedrop in dielectric permittivity, De, around the criticalfrequency fC (Fig. 1). Assuming that each cell is surroundedby an intact membrane (m) of small thickness dwith constantsurface capacitance cm¼ em/d and negligible transconduc-tance gm¼ sm/d� 0, for a dilute suspension the drop inpermittivity is (Foster and Schwan, 1989; Maxwell, 1881;Schwan, 1957)

De � 9cm4

pR: ð1ÞSince DC/De, linear dependence of De on the mathemat-

ical product p�R allows biovolume to be monitored usingdifferences in bulk suspension capacitance for two frequen-cies from the opposite ends of the range 0.1–20MHz (Harriset al., 1987; Opel et al., 2010). Since volume fraction, p, is theratio of the total volume of N suspended cells of averagevolume Vcell¼ 4pR3/3 to the volume of suspension, Vtotal,

Equation (1) results in

De / NR4: ð2ÞThis results in the expectation that De should only depend

on the number of cells, N, and their average radius, R.However, this simplified relationship has been shown todeviate when there are changes in the physiological or meta-bolic state of mammalian cells as they lose viability in thestationary anddeclinephasesof abioprocess (Opel et al., 2010).In recent years, dielectric spectroscopy has been used

successfully for cell viability measurements with the advan-tage of constant monitoring that could enable early correctiveaction to maximize cell viability (Ducommun et al., 2002).The prototype DEP cytometer developed at the University ofManitoba (Nikolic-Jaric et al., 2013) probes dielectricproperties of individual cells by independently detectingand actuating the cells as they flow through the microfluidicchannel. Cells respond to the applied electric field bypolarization, which can be defined in terms of variations incharge distribution. As each cell flows over the electrodearray, a change in capacitance can be registered. This providesinformation of the dielectric properties of individual cells in apopulation as opposed to the bulk average dielectricproperties obtained with an online capacitance probe.Mostmammaliancells inabioprocess loseviabilityasa result

of programmed cell death, or apoptosis (Kerr et al., 1972),initiated by a variety of triggers such as oxidative stress,starvation, and ER stress (Limoli et al., 1998; Rao et al., 2006;Simon and Karim, 2002). Even though the cell membrane isintact at the onset of apoptosis, once a cell commits to theprogrammed death the process cannot be reversed. Therefore,it is reasonable to include all cells consigned to apoptosis into anon-viable count. With the aim of increasing volumetricproductivity and maintaining a constant product quality,indicators of early events leading to apoptosis are highly

Figure 1. b-dispersion, schematic plot. Charge accumulation on the membranes

occurs over a finite amount of time, resulting in a large drop in dielectric permittivity,De(from es at low to e1 at high frequencies) at the critical frequency, fC. Values are plotted

for dimensionless frequency, f/fC.

Braasch et al.: The Changing Dielectric Properties of CHO Cells 2903

Biotechnology and Bioengineering 2903

desirable in cell bioprocesses because these initial stagesmaybereversible by appropriate intervention—for example, nutrientfeeding (Geske et al., 2001). While themid-stages of apoptosisindicate the end of production from the cells, late stages maycause harm as lysed cells release proteinases and glycosidasesinto the medium with potential to degrade the glycoproteinproduct (Winchester, 2005).

One of the earliest indications of apoptosis in amammalian cell is the presentation of phosphatidyl-serineon the cell surface (Martin, 1995). In addition cells shrink(Bortner and Cidlowski, 2002, 2003; Kerr et al., 1972) due toionic cell content regulation (Panayiotidis et al., 2006) andchromatin condensation (Kerr et al., 1972). Early pro-apoptotic signals trigger a cascade of caspase activity withinthe cells (Budihardjo et al., 1999). Finally, the late stage ofapoptosis is characterized by DNA fragmentation and loss ofmembrane integrity. The phases of apoptosis can be detectedby offline measurements of a sample from a bioreactor usingfluorescent markers with a flow cytometer. Specific markersare available for early, mid and late stages of apoptosis.

A number of studies have connected the ionic flux over thecell membrane with the onset of programmed cell death(Bortner and Cidlowski, 2002, 2003; Bortner et al., 2012) andfound that changes in ioniccompositionof the cell are linked toearly stages of apoptosis (Labeed et al., 2006). However, theheterogeneity of the growth cycle phase of individual cellsobscures the detection of these changes when using bulkcapacitance measurements, which show average dielectricproperties of the cell suspension. The phenomenon ofdielectrophoresis (DEP), in which the uncharged particlesmove along the gradient of a non-uniform electric field,provides a means for monitoring dielectric changes inindividual cells. These changes correlate with physiologicaland metabolic changes (Demierre et al., 2008; Duncanet al., 2008; Flanagan et al., 2008; Gagnon, 2011; Huanget al., 1992; Pethig et al., 2010). Therefore, methods based onDEP are particularly promising for identifying emergingsubpopulations of apoptotic cells.

The first objective of this research was to determine howmeasurements from an online capacitance probe comparedto those from three other commonly used methods (Cedex,Coulter Counter, ViaCount flow cytometer assay) for celldensity determination. The second objective was to deter-mine the loss of cell viability by dielectric spectroscopicanalysis of individual cells using a prototype DEP cytometerin comparison to three independent methods (Cedex,ViaCount and Nexin flow cytometer assays). We show thatthe DEP cytometer could detect the dynamic shift ofsubpopulations of cells with varying dielectric propertiescoinciding with loss of cell viability at the end of a bioprocess.

Materials and Methods

Cell Culture and Sampling

Chinese Hamster Ovary (CHO) cells expressing a human-llama chimeric antibody (EG2) for epidermal growth factor

receptor (EGFR) were used throughout this work. The cellline (CHODG44-EG2-hFc/clone 1A7) was kindly providedby Yves Durocher of the NRC, Canada (Bell et al., 2010). Thecells were cultured in a 3 L glass bench-top bioreactor(Applikon, Foster City, CA) with the following set-pointsmaintained throughout: 37�C, 30% dO2, 7.2 pH, and200 rpm. The cells were grown in BioGro-CHO serum-freemedium (BioGro Technologies, Winnipeg, MB) supple-mented with 0.5 g/L yeast extract (BD, Sparks, MD), 1mMglutamine (Sigma, St. Louis, MO), and 4mM GlutaMax I(Invitrogen, Grand Island, NY). Two samples (10mL each)were taken from the cell culture at regular intervals tomonitor cell density and viability. Each sample was analyzedtwice using the Cedex, Coulter Counter, and Guava Assaysgiving a total of four technical replicates at each samplingpoint for these offline measurements. The capacitance probewas a continuous online measurement and two technicalreplicate measurements were performed at each samplingpoint using the DEP Cytometer.

Cell Density and Viability Monitoring

Trypan Blue Exclusion

A Cedex Image Analyzer (Innovatis AG, Bielefeld, Germany)was used to analyze cell samples diluted 1:2 with 0.4% (w/v)trypan blue (Invitrogen) to determine cell density andviability. The Cedex was calibrated using polystyrene ballssimilar in size to CHO cells (standard calibration method). Inaddition, a light calibration (to correct exposure) wasperformed every 3–4 months. A comparison of viable cellcounts by Cedex and a hemocytometer method using trypanblue exclusion gave values within 5%.

Coulter Counter

Cell samples were diluted 1:125 with PBS (Invitrogen)containing 1mM EDTA (Fisher, Fair Lawn, NJ) and analyzedusing a Coulter Counter (Coulter Electronics, Inc., Hialeah,FL) to determine total cell density.

Flow Cytometer

Three different assays were performed using the Guava 8HTsystem and easyCyte software (EMD Millipore, Danvers,MA). Initially, a ViaCount (Catalog No. 4000-0040, EMDMillipore) assay was performed to determine the viability andcell density during the bioprocess. The ViaCount assayconsists of two DNA binding dyes—one that enters all cellsand the second that only stains cells with a compromisedmembrane. The assay was performed based on the suppliers’protocol with an incubation time of 10min. To achieve an in-depth analysis of the culture Nexin and caspase-8 (EMDMillipore) assays were also performed. The Nexin assay(Catalog No. 4500-0450) is based on the detection ofphosphatidyl serine on the outer membrane of cells, acharacteristic of early apoptosis. The assay was performed

2904 Biotechnology and Bioengineering, Vol. 110, No. 11, November, 2013

using the suppliers’ protocol with 200mL of Guava NexinReagent (EMDMillipore) and an incubation time of 20min.The caspase-8 assay (Catalog No. 4500-0550) is based on thedetection of the active caspase 8 enzyme and was performedaccording to the supplier’s protocol. The 7-amino-actinomy-cin D (7-AAD) stain was incorporated within the Nexin andcaspase-8 assays to determine cells in late stage apoptosis andthese were designated as “dead”. Designation of sub-populations of cells following the Nexin and caspase-8 assayswas based on the acquisition of 2000 data points per samplefrom the flow cytometer. The quadrant markers werepositioned manually following the supplier’s protocol.Figure 2 shows a representative image with viable cellspositioned in the lower left quadrant.

Capacitance Measurement

For the on-line capacitance measurement a capacitance probe(Aber Instruments, Aberystwyth, UK) was incorporatedaseptically into the bioreactor. The probe was connected viaan amplifier to a 220 Model detector and set to measure DCthroughout the bioprocess. The probe was set to perform afrequency sweep from 0.1 to 20MHz every minute.

DEP Cytometer

In the prototype DEP cytometer designed at the University ofManitoba, cells are actuated by a DEP force resulting from the

AC signal, with a frequency from 0.1 to 6MHz, appliedthrough the set of actuation electrodes (Nikolic-Jaricet al., 2013) (Fig. 3). This results in a vertical displacementof the cells between the two detection sites (D1 andD2) and istranslated into an electronic signature, S, as shown in Figure 3(a). The peaks, P1 and P2, depend on the vertical position ofthe cells’ center of mass and diminish with the distance fromthe electrodes. Vertical cell displacements as small as 0.1 to0.25mm can be registered, permitting determination of theextent of actuation that occurs between sites D1 and D2 bysimply comparing P1 and P2.As shown in Supplementary material, DEP force is

proportional to cell polarizability, a. The force is attractive(pDEP), fora> 0,andrepulsive (nDEP)fora< 0;pointa¼ 0 iscalled the “crossover.” Numerical simulations of a/(3 esm),where esm is therealpermittivityof thesuspendingmedium,areshown in Figure 3(a). They are based on the assumption thatthe cell shrinks while at the same time conductivity ofcytoplasm changes as a result of the intracellular ion fluxesstimulated by early apoptotic changes (Bortner andCidlowski, 2003). Simulations estimate the crossover frequen-cy(whena¼ 0) for viablecells atabout0.5MHz.Therefore,ACDEP signals above 1MHz can distinguish between the healthyviable and early apoptotic cells, as the difference inpolarizability will produce DEP forces of opposite sign.We used a force index f to quantify the extent and

direction of the cell actuation:

f � P2 � P1P1 þ P2

: ð3Þ

Note that f¼ 0 represents no actuation, with the cellremaining at the same elevation (this occurs either when noDEP force is applied or when the cell is dielectricallyindistinguishable from the suspending medium); f> 0 orf< 0, indicates a cell that experiences a pDEP or an nDEPforce, respectively—cf. Figure 3(a).For the offline measurement a sample (5–10mL) was

taken from the bioreactor at timed intervals, centrifuged(377g) and the cell pellet reconstituted in fresh growthmedium (37�C) and a low conductivity (�0.067 S/m)medium (37�C) [22.9mM sucrose (Sigma), 16mM glucose(Fisher), 1mM CaCl2 (Fisher), 16mM Na2HPO4 (Fisher)](Polevaya et al., 1999). The cell suspensions were diluted to105 cells/mL using a ratio of 2:30 (fresh: low conductivitymedium) to reach a conductivity of �0.17 S/m as measuredby a conductivity meter (Orion 3-Star Plus, ThermoScientific, Waltham, MA). An initial time course showedthat cell viability and cell size were stable (within 5% of theoriginal sample) for at least an hour—the time needed for theDEP measurement during the experiment.

Statistics

Differences between measurements were assessed by un-paired two-tailed Student’s t-test. A P-value of <0.02 wasconsidered as statistically significant.

Figure 2. Example of CHO EG-2 cells assayed using the Guava Nexin assay.

Before acquisition the voltages were set to position the viable cells in the lower left

corner of the Annexin V-PE versus 7-AAD dot plot. During analysis the quadrant

markers were manually positioned to separate the distinct populations of cells. This

allows identification of the different apoptotic stages identified by the Nexin assay: I

viable/non-apoptotic cells [Annexin V-PE (�) and 7-AAD (�)]; II early-apoptotic cells

[Annexin V-PE (þ) and 7-AAD (�)]; III late stage apoptotic/dead cells [Annexin V-PE

(þ) and 7-AAD (þ)]; IV nuclear debris [Annexin V-PE (�) and 7-AAD (þ)].



Results

Cell Culture profiles

The results of the comparative measurements of cell densityand viability for two separate batch cultures (A inoculatedfrom passage 25, B inoculated from passage 35) are shown inFigures 4 and 5. Profiles in panel A of both Figures 4 and 5were based on samples taken every 24 h from inoculation upto 170 h. In a second culture (panel B) samples were analyzedevery 6 h from 96 to 130 h in order to monitor the viable tonon-viable transition of the cells more closely. Total andviable cell densities (Fig. 4) were determined using fourdifferent methods: Coulter counter, Cedex, ViaCount and insitu capacitance probe. The data in Figure 4 show that duringthe exponential growth phase each set of measurements gavesimilar values. However, divergence in measurementsoccurred beyond the point of maximum cell density at 90–

100 h, which coincided with a decline in cell viability as seenin Figure 5. The Coulter counter provides an electronicmeasure of presence of particles and does not distinguishbetween viable and non-viable cells; consequently, high celldensity values were observed up to the end of the cultures.The other threemeasurements showed an apparent decline incell density beyond the point of maximum cell density.However, there was a statistically significant difference(P< 0.02) between the Cedex measurements and themeasurements by the capacitance probe (4A after 120 h, 4Bafter 100 h) and the ViaCount assay (4A after 120 h, 4B after100 h). The cell density determined by the capacitance probewas comparable to measurements made using the ViaCountassay. The consistently higher values determined from theCedex can be explained by the mechanism of trypan dyeexclusion, which depends upon cell membrane damage—alate event in apoptosis (Hughes et al., 1997). Although theViaCount assay is based on the same principle of dye

Figure 3. a: Schematic representation of the microfluidic channel. A cell (modeled to scale as a small sphere) is flowing through the analysis volume above the differential

electrode array. Three possible cell trajectories are shown: pDEP-actuated cells are attracted to the electrodes, nDEP-actuated cells are repelled, and unactuated cells continue

along a straight line. Corresponding signatures (in the vertical plane) are the true signatures produced by CHO cells of similar sizes captured at different times during the experiment.

Similarly sized entrance peak, P1, indicates that they were initially flowing at approximately the same elevation. The change in amplitude resulted from about 4mm vertical deflection

in both the pDEP and nDEP examples. b: Micrograph of the differential electrodes, showing detection and actuation region, with a CHO cell on the exit from the analysis volume.

Figure 4. Comparison of cell density and capacitance determined for two representative CHO EG2 batch cultures using Trypan blue (Cedex), Coulter Counter, Guava flow

cytometer (ViaCount) and Capacitance probe (Aber). The culture (3 L) was sampled every 24 h from 0 to 168 h (A). The culture (3 L) was sampled every 24 h from 0 to 95 h and then

every 6 h from 95 to 119 h (B). The continuous DCmeasurement taken from the probe was adjusted to align with cell density at 72 h determined by Cedex. The values graphed for the

off-line measurements are the average of four technical replicates for each sampling point with error bars representing the standard error of the mean.


exclusion it appears to be able to detect an earlier stage ofmembrane damage and hence registers lower viabilitiescompared to trypan blue exclusion (Millipore TechnicalPublication: MK80000410).Figure 5 shows data from the same two bioreactor cultures

monitored for the percentage viability of the cell population.As in the previous data this shows a decline in viabilitybeyond the maximum cell density at 90–100 h. Divergencecan be observed for the rate of loss of viability as determinedusing the Cedex (trypan blue), ViaCount assay, Nexin assay,and DEP cytometer. Statistical analysis shows that the Cedexviability determination was significantly different (P< 0.02)from measurements made by the ViaCount assay (5A after70 h, 5B after 75 h), the Nexin assay (5A after 70 h, 5B after90 h), and the DEP cytometer measurements (5A after 100 h,

5B after 100 h). In addition measurements by the ViaCountassay were significantly different (P< 0.02) from measure-ments made by the Nexin assay (5A between 100 and 145 h,5B after 90 h) and the DEP cytometer measurements (5Abetween 100 and 145 h, 5B after 105 h). Themeasurements bythe Nexin assay and the DEP cytometer were not significantlydifferent before 118 h of culture. However, there was asignificant difference (P< 0.02) between the measurementsafter that time for both runs. These profiles are consistentbetween the two cultures shown in Figure 5 and otherrepeated runs (not shown).In addition to the above measurements the glucose and

lactate concentrations were also monitored for bioreactorruns A (passage 25) and B (passage 35) (Fig. 6a). The data inFigure 6a show that for both runs the start of the decline

Figure 5. Comparison of cell viability determined for CHO EG2 batch cultures using trypan blue (Cedex), ViaCount and Nexin (Guava flow cytometer) and DEP cytometer. The

culture (3 L) was sampled every 24 h from 0 to 168 h (A). The culture (3 L) was sampled every 24 h from 0 to 95 h and then every 6 h from 95 to 119 h (B). The values graphed are the

average of four technical replicates for each sampling point with error bars representing the standard error of the mean. Each DEP value is based on two determinations from>200

cells per sample (error bars represent 5%).

Figure 6. a: Comparison of glucose consumption and lactate accumulation for run A (passage 25) (&) and run B (passage 35) (~). Concentrations of glucose (---) and lactate

(—) were plotted against days in culture. Each point represents the average of two technical replicates with error bars representing the standard error of the mean. b: Comparison of

the average yield lactate/glucose (Ylac/glc) for run A (passage 25) and run B (passage 35) during the exponential growth phase. Each bar represents the average of two technical

replicates with error bars representing the standard error of the mean (�P< 0.02).



phase in Figure 4 A and B correlates with the glucoseconcentration in the medium dropping below 1mM (run Aat �100 h and run B at �90 h). Furthermore in run Bbetween 50 and 95 h glucose was consumed slightly faster andlactate accumulated more than in run A. This difference inlactate production during the run in combination with adifference in glucose consumptionmanifests in a significantlyhigher yield of lactate per glucose in the exponential growthphase of run B (Fig. 6b) (P< 0.02).

Overall, the profiles of the two bioreactor runs were similaralthough the time difference in the onset of apoptosis in runA (�100 h) and B (�90 h) (Figs. 4 or 5) may be attributed tothe difference in the metabolic state of their respectiveinoculum and the subsequent difference in metabolismduring the run (Fig. 6).

Measurements and Modeling With the CapacitanceProbe

Figure 4 shows that the profile of the bulk capacitancemeasurement from the online probe in the cultures followsthe ViaCount measurements fairly closely during theexponential phase and indicates the decline of the viablecell population. Some divergence was detected at the point inwhich non-viable cells predominate as shown at 140 h inFigure 4A. Closer examination of the profile from thecapacitance probe shows a distinct inflexion at 140 h forculture A and 120 h in culture B (Fig. 7).

We examined the divergence of these profiles in relation tothe simplified theoretical Equations (1) and (2). As ourexperimental observations show that mean cell radius, R,remains at a constant value, Rv, during the growth phase of6.5mm as measured by Cedex, De increases proportionallyto the increase in the N, reaching the value ofDemax ¼ NR4v at

�100 h for run A, and �90 h for run B (Fig. 4). At Demax, atotal of N cells are assumed to be healthy (non-apoptotic),with an average radius Rv. Beyond Demax, as is evident fromthe Coulter counter measurements (Fig. 4), N remainsconsistently high even as the bioprocess population under-goes apoptosis and De steadily declines. With N� const, itfollows from Equation (2) that the decrease in De can only bedue to changes in the average cell radius. This hypothesis canbe explicitly tested by analyzing the bulk capacitance dataobtained by the probe at 0.3MHz. Even though the trypanblue viability estimates begin to decline after Demax, theyremain above 70% until�145 h (run A) and above 80% until�120 h (run B), the times when DEP cytometer and Nexinassay both indicate that the predominant cell population isapoptotic. Assuming that the average radius of the viable cellsremains at Rv while the radius of apoptotic cells changes torRv. From the data on the average decrease in size of all cellswith intact membranes obtained by Cedex Image analysis(not shown) we estimate the average decrease in size ofapoptotic cells as r� 0.87. For example, data from culture Aindicate that themean cell diameter changed from 13.2mmat24 h to 11.5mm at 168 h when most cells were apoptotic. TheDEP cytometer identifies two subpopulations, viable andapoptotic, contributing in fractions v and (1–v), respectively.This datawas used tomodel the expected drop in permittivityas (see Supplemental material for derivation).

De ¼ ½v þ ð1� vÞr4Demax: ð4ÞThe bulk capacitance measurements and the model are

both represented in Figure 7. The divergence between thetheoretical and observedDe increases progressively over time,resulting in a considerable discrepancy towards the end of theculture. Even if the average radius of all cells were to decreaseto 0.87 Rv, De would not be expected to decrease below 0.57

Figure 7. Bulk capacitance probe profiles compared to values predicted by cell volume changes. The continuous lines represent experimentally determined profiles from the

two cultures described in Figure 4. Individual points (O) represent the model predictions for a decrease inDe if two subpopulations are present as determined by DEP cytometer. Forthe model, the assumed change in radius of the apoptotic cells was 13% (R! 0.87Rv) based on CEDEX measurement. It is assumed that the radius of viable cells does not change.


Demax. However, the value of run A measurements at 145 hand v¼ 0.16 is De¼ 0.454 Demax, and the value of run Bmeasurements at 120 h and v¼ 0.09 is De¼ 0.415 Demax,both well below the theoretically predicted values. Thus, weconclude that the decrease in De during the decline phasecannot be explained only by the decrease in average cellradius, but should instead be attributed to the decrease inboth the cell radius and the cell polarizability of a dielectri-cally diverse subpopulations of cells. The decrease inpolarizability could be due to changes in cell metabolism,membrane surface capacitance, and conductance, possibly bythickening of cell membrane or changed cytoplasmicconductivity.

Subpopulations of Cells Analyzed by Flow Cytometry

Figure 8 shows the cell subpopulations found in samples fromthe two separate bioreactor cultures as measured by the flowcytometer. Up to 70 h only the ViaCount assay was used todifferentiate between the viable and non-viable cells. After70 h a more in-depth analysis of the cell populations in thebioreactor cultures is shown by identifying and quantifyingthe non-apoptotic, early-apoptotic, caspase 8 positive, anddead subpopulations (Fig. 8) using the Nexin and Caspase 8assay. The bar graphs in Figure 8 clearly show that the cellpopulation in the bioreactor is not homogeneous and thatnot all cells progress through apoptosis at the same time. Inaddition, Figure 8 shows the increase in size of the earlyapoptotic sub-population over time, with eventual transitionto late apoptotic stage. Figure 8 also shows a consistently lowfraction of caspase-8 (þ) and 7-AAD (�) cells after 70 h forculture A and B. Over the course of the bioprocess the totalfraction of cells staining caspase-8 (þ) increased steadily untilmost cells stained caspase-8 (þ) at the end of the run. The

majority of these cells were late stage apoptotic and dying/dead cells staining caspase-8 (þ) and 7-AAD (þ) (data notshown).

DEP Cytometer Measurements

Examples of experimental DEP signatures produced byindividual CHO cells are shown in Figure 3(a). In asuspension with a conductivity of ssm� 0.17 S/m thecrossover frequency was identified experimentally as�0.5MHz above which viable cells produced pDEPsignatures and non-viable cells nDEP signatures. The forceindex was measured for CHO cells in nine discrete frequencysteps, from 0.1 to 6MHz as identified in Figure 9 andsimulated based on parameters in Table I. Experimentalsignatures from these samples were collected for 200–400cells (�5min) at each frequency and each control set. Figure9(b) represents stacked distributions of force index values foreach of the nine discrete steps during the frequency sweep.These distributions were obtained from two samples takenearly in the culture, and after the decline phase. Viability wasdefined as the fraction of pDEP signatures in the total numberof cell signatures captured. DEP cytometer viability countsfor these samples were 93% and 9%, respectively. Duringrepeated runs we were able to show that the percent viabilitydetermined by the DEP Cytometer was consistent with thedata from the Nexin assay (Fig. 5).A detailed analysis of the data from the DEP cytometer

during the critical time in which cells display a rapid loss inviability shows the changes in force index distribution for thecell population (Fig. 10). Within this time interval (96–120 hfrom inoculation in culture B), signatures from 300–400 cellsactuated with 6MHz DEP signals were collected every sixhours (only data at 12 h intervals are shown). Each of the

Figure 8. Overview of subpopulations determined during the bioprocess using three flow cytometer assays (ViaCount, Nexin, caspase 8). The culture (3 L) was sampled every

24 h from 0 to 168 h (A). The culture (3 L) was sampled every 24 h from 0 to 95 h and then every 6 h from 95 to 119 h (B). During exponential growth cells were distinguished as viable

( ) or dead ( ) based on the ViaCount assay. During the late exponential and the decline phase (>70 h) the cell subpopulations were distinguished as non-apoptotic (Annexin V (�)

and 7-AAD (�)) ( ), early apoptotic (Annexin V (þ) and 7-AAD (�)) (&), caspase 8 (caspase 8 (þ) and 7-AAD (�)) (&), or dead (7-AAD (þ)) ( ). The values graphed are the average

of mostly four technical replicates for each sampling point. The proportion of dead cells determined by the ViaCount assay was within <10% of the value determined by the Nexin

and caspase 8 assays.



resulting histograms shows a sharply bimodal distribution,with a clear shift over time from predominantly positive topredominantly negative force index values. These results areconsistent with the data from the Nexin assay, illustrated inFigures 5 and 8. The table in Figure 10 shows that the nDEPmeasurements determined over the period of changing cellviability has a closer correlation to the Nexin assay than theother colorimetric or fluorescent assays used. This suggeststhat the DEP measurements are indicative of changes relatedto the early stages of the apoptotic process.

Discussion

The ability to prolong the viability of cells at high density is akey factor in enhancing the productivity of a mammalian cell

bioprocess (Wurm, 2004). This requires an understandingand monitoring of the events that take place during thedecline phase and loss of cell viability, which is usuallyassociated with apoptosis. Viability has often been defined bythe loss of integrity of the cell membrane which allows theentry of high molecular weight dyes like trypan blue(Strober, 2001). However, this is likely to be a late stageevent in the demise of CHO cells in a bioreactor that undergoprogrammed cell death (Browne and Al-Rubeai, 2011).During apoptosis the membrane can remain intact duringevents of cell shrinkage, chromatin condensation and caspaseactivation (Hughes et al., 1997). An early molecular changeduring apoptosis is the translocation of the phosphatidylserine to the outer leaf of the membrane and the activation ofa series of caspases (Budihardjo et al., 1999; Martin et al.,

Figure 9. a: Numerical simulation of the polarizability, produced for a double-shell model (Jones, 1995) and based on the assumption that cell volume shrinks and the

conductivity of the cytoplasm drops as the cell undergoes the apoptotic process; parameters used in this simulation were obtained from previously published work (Asami

et al., 1989; Cemazar et al., 2011) as given in Table I (Refs. 2 and 1, respectively). This simulation was used as an aid in choosing 6MHz as the frequency at which to perform the

viability measurements. b: Stacked distributions of force index, f, for the collection of signatures captured early and late in a previous experiment (Days 1 and 10, respectively).

Signatures were collected at nine discrete steps in frequency. Two collections were made at 6MHz frequency at the beginning and at the end (6B and 6A standing for before and

after) of the time interval to do the full frequency sweep and to ensure that the viability of cells was not changed during the course of the experiment.

Table I. List of parameters used to numerically simulate the polarizability of CHO cells according to the double-shell model (Jones, 1995).

Parameter Viable Non-viable Units Source

GeometryCell radius 6.50 5.53 mm MeasuredNuclear radius 3.25 mm EstimatedMembrane thickness 5 4.25 nm Ref. 1Nuclear envelope thickness 40 nm Ref. 1

ConductivitySuspending medium 0.17 S/m MeasuredCytoplasm 0.32 0.17 S/m Ref. 2Cell membrane 3 15 mS/m Ref. 2Nucleoplasm 1.35 0.17 S/m Ref. 2Nuclear envelope 6 30 mS/m Ref. 2

Relative permittivity (vacuum permittivity, e0¼ 8.85� 10�12 F/m)Suspending medium 78 e0 Ref. 2Cytoplasm 60 e0 Ref. 2Cell membrane 6.80 e0 Ref. 2Nucleoplasm 52 e0 Ref. 2Nuclear envelope 28 e0 Ref. 2

Refs. 1 and 2 refer to Cemazar et al. (2011) and Asami et al. (1989), respectively.


1995). These can be labeled by fluorescent stains andmonitored in a cell population via flow cytometry. Thisbrings into question the most appropriate measure ofviability, as a population of cells undergoing apoptosis mayindicate significantly different values of percentage of viabilitydepending on the assay used.The profiles of the two bioreactor runs in this study were

similar with the second run showing faster growth, increasedglucose uptake and lactate accumulation during theexponential growth phase. This in turn led to an earlierglucose depletion and subsequent earlier onset of apoptosisattributed to nutrient depletion in the culture. The differencein the observed metabolism may be attributed to the use ofcells from different passage numbers for inoculation. Thisfinding agrees with observations made by Beckmann et al.(2012) for cells of higher passage numbers. However, overalltrends between methods used and the onset of apoptosis inaccordance to nutrient (i.e., glucose) depletion are the samefor both runs.Our study compared four different methods to monitor

and evaluate the changes in viable cell densities and cellviability that occur to CHO cells at the later stages of aculture. The electronic counter (Coulter) is a rapidmethod ofdetermining cell density but does not distinguish betweenviable and non-viable cells. Trypan blue exclusion is a widelyused colorimetric method that is usually combined withhemocytometer counting or image analysis and distinguishes

the color of non-viable cells (blue) and viable cells (non-colored). However, as we and others have shown thisoverestimates viability by not recognizing apoptotic cells withan intact membrane (Altman et al., 1993; Browne and Al-Rubeai, 2011) but is rather dependent on gross membranedamage, which occurs at a later stage of apoptosis (Koopmanet al., 1994).The flow cytometer analysis can be combined with a range

of florescent dyes to determine different stages of apoptosisincluding phosphatidyl serine translocation (Nexin), cellmembrane permeability (ViaCount), and caspase activation(e.g., caspase 8). The Nexin assay recognizes an early event inapoptosis and so it is not surprising that our results indicatethat among the fluorescence assays it produces the lowestestimates of viability. The caspase-8 assay was used torecognize cells positive for caspase-8 activation indicatingcells going through the extrinsic cell surface death receptorapoptotic pathway (Budihardjo et al., 1999). While only asmall fraction of cells was determined to be caspase-8 (þ) and7-AAD (�) at each sampling point, the overall fraction ofcaspase-8 (þ) and 7-AAD (þ) cells increased over the time ofthe bioprocess. This suggests that in most cells the caspase-8initiator caspase was activated but that the transition fromcaspase-8 (þ) and 7-AAD (�) to caspase-8 (þ) and 7-AAD(þ) was too rapid to detect. With the caspase-8 assay used weobserved up to 15% of the cells in the bioprocess to benecrotic for some sampling points. To determine whether this

Figure 10. Time-lapse histograms showing a binary population of cells with rapidly changing dielectric properties (�96–120 h from bioreactor seeding) accompanied by a

comparison table of viability assessments for different cell monitoring techniques.



percentage was an artifact of the staining procedure a visualassessment of the Nexin assay would have to be performed todetermine if phosphatidyl serine was found on the insideof Annexin V (þ) and 7-AAD (þ) cells (Wlodkowicet al., 2011). This would be indicative of a necrotic cell.

For future experiments it would be interesting to use thecaspase-3/7 assay to monitor the activation of the key effectorcaspase 3 during the bioprocess (Vermes et al., 2000), which isactivated by both caspase 8 and 9. However, it was interestingto note a steady increase of Caspase 8 (þ) staining cells duringthe cultures as this agrees with findings that a depletion inglucose is a potential cause for caspase 8 activation (Muñoz-Pinedo et al., 2003; Yun et al., 2007).

The capacitance probe has been widely accepted in thebiotechnology industry as a sterilizable probe that can beinserted in situ in a bioreactor for continuous monitoring ofviable cell densities (Carvell and Dowd, 2006). Previousreports have confirmed our conclusion that on-line measure-ments for biomass by capacitance correlate well with thetrypan blue assay of viable cell concentration (VCC) in theexponential phase of cell growth but deviate in eitherdirection in the stationary and the decline phases (Ducom-mun et al., 2002; Opel et al., 2010).

The changing capacitance measurements are based on thetotal biovolume as well as the membrane charge on each cell.These parameters are likely to be constant during exponentialgrowth of cells. However, interpretation of the data becomesmore difficult during the stationary and decline phases forwhich our data indicates a non-homogeneous culture inwhich reduction in both the cell size and surface charge occurat varying degrees. The data we measured using the change ofDe and the software interpretation from the capacitanceprobe correlates well with the ViaCount fluorescent assayanalyzed using the Guava flow cytometer and provides amorerealistic measure of viable cell density in the final stages ofculture compared to trypan blue exclusion. However,deviations from the simple model of a direct relationshipbetween capacitance and biovolume during the decline phasecould be attributed to a decrease in both the cell size andpolarizability as cells become apoptotic. This observation iscompatible with previously reported changes due to alteredmetabolism in the decline phase (Opel et al., 2010).

In this study we describe a novel DEP cytometer thatenhances the analysis of cells during the critical stage ofculture as cell viability decreases. The DEP-derived data isbased on individual cells flowing over a detector and thereforediffers from the averaged bulk dielectric measurements madewith the capacitance probe. The DEP cytometer measures thedielectric properties of individual cells and was able toidentify at least two populations of cells, each with a distinctpolarizability as distinguished by the measured force index.One population was associated with viable cells and the otherwith apoptotic cells. From the end of the exponential throughthe stationary and decline stages there was a gradual shift ofcell count from the viable into the apoptotic population;however, the two populations always maintained theirindividual dielectric properties throughout this shift.

Substrate deprivation during the later stages of a culture canlead to declining intracellular concentrations of ATP(Burgener et al., 2006) thatwould in turn lead toperturbationsin the membrane-bound Naþ/Kþ ATPase essential formaintaining resting electrical potential of cells (Jaitovichand Bertorello, 2006). The appearance of subpopulations ofcells, which results in the changes in dielectric properties asanalyzedby theDEPcytometer, canbeattributable to the rapidchange in cell conductivities, resulting from ionic efflux orinflux thatmayact as an“on” switch for apoptosis, as suggestedby earlier literature (Bortner and Cidlowski, 2002, 2003).

An advantage of the DEP cytometry is that it can be used asa non-invasive diagnostic and research tool that does notrequire any specificmarkers or dyes. The DEP cytometer usedin this work is a prototype with a promise to become anelectronic monitor of physiological changes in the cell.Furthermore, the all-electronic operation principle makesthis instrument particularly suitable for miniaturization andfully automated analysis.

Notations

7-AAD 7-amino-actinomycin D.DEP dielectrophoresis (pDEP, nDEP—positive,

negative DEP)~e complex dielectric permittivity" real dielectric permittivity"0 dielectric permittivity of vacuums conductivityv (radial) frequency of the AC electric field;

v¼ 2pf, where f is ordinary frequencyD" change (drop) in real dielectric permit-

tivity due to b-dispersionD"max maximum value of D"fC critical frequency at which the drop in

permittivity is D"/2cm surface capacitance of the cell membranegm transconductance of the cell membraned cell membrane thicknessDC change in capacitanceN number of cells; total number of cells in a

bioreactorVcell average cell volume; Vcell ¼ 4pR3=3,

where R is the average cell radiusVtotal total volume of the suspensionp¼NVcell/Vtotal volume fraction of the total of N cells in

suspensiona polarizability"sm, ssm permittivity and conductivity of the

suspending mediumS electronic cell signatureP1, P2 peaks in signature Sf force indexrR radius of the cell after decrease; r 1v fraction of viable (non-apoptotic) cells, as

determined by DEP cytometer measure-ments


The authors would like to thank Aber Instruments and EMDMillipore for equipment loan. We also acknowledge NSERC, themonoclonal antibody network (MabNet) as well as the CanadaFoundation for Innovation (CFI); the Province of Manitoba; WesternEconomic Diversification Canada (WD); and Canadian Microelec-tronics Corporation (CMC) Microsystems for financial support ofthis research. In addition, the authors thank John Carvell and BobTodd for review of the manuscript and valuable discussion.

References

Altman SA, Randers L, Rao G. 1993. Comparison of trypan blue dyeexclusion and fluorometric assays for mammalian cell viabilitydeterminations. Biotechnol Prog 9:671–674.

Ansorge S, Esteban G, Schmid G. 2010. On-line monitoring of responses tonutrient feed additions by multi-frequency permittivity measurementsin fed-batch cultivations of CHO cells. Cytotechnology 62:121–132.

Asami K, Takahashi Y, Takashima S. 1989. Dielectric properties of mouselymphocytes and erythrocytes. Biochim Biophys Acta 1010:49–55.

Bebbington CR, Renner G, Thomson S, King D, Abrams D, Yarranton GT.1992. High-level expression of a recombinant antibody from myelomacells using a glutamine synthetase gene as an amplifiable selectablemarker. Nat Biotechnol 10:169–175.

Beckmann TF, Krämer O, Klausing S, Heinrich T, Thüte T, Büntemeyer H,Hoffrogge R, Noll T. 2012. Effects of high passage cultivation on CHOcells: A global analysis. Appl Microbiol Biotechnol 94:659–671.

Bell A, Wang ZJ, Arbabi-Ghahroudi M, Chang TA, Durocher Y, Trojahn U,Baardsnes J, Jaramillo ML, Li S, Baral TN, O’Connor-McCourt M,Mackenzie R, Zhang J. 2010. Differential tumor-targeting abilities ofthree single-domain antibody formats. Cancer Lett 289:81–90.

Bortner CD, Cidlowski JA. 2003. Uncoupling cell shrinkage from apoptosisreveals that Naþ influx is required for volume loss during programmedcell death. J Biol Chem 278:39176–39184.

Bortner CD, Scoltock AB, Sifre MI, Cidlowski JA. 2012. Osmotic stressresistance imparts acquired anti-apoptotic mechanisms in lymphocytes.J Biol Chem 287:6284–6295.

Bortner CD, Cidlowski JA. 2002. Apoptotic volume decrease an theincredible shrinking cell. Cell Death Differ 9:1307–1310.

Browne SM, Al-Rubeai M. 2011. Defining viability in mammalian cellcultures. Biotechnol Lett 33:1749.

Budihardjo I, Oliver H, Lutter M, Luo X, Wang X. 1999. Biochemicalpathways of caspase activation during apoptosis. Annu Rev Cell Dev Biol15:269–2290.

Burgener A, Coombs K, ButlerM. 2006. Intracellular ATP and total adenylateconcentrations are critical predictors of reovirus productivity from Verocells. Biotechnol Bioeng 94:667–679.

Butler M. 2005. Animal cell cultures: Recent achievements and perspectivesin the production of biopharmaceuticals. Appl Microbiol Biotechnol68:283–291.

Carvell JP, Dowd JE. 2006. On-line measurements and control of viable celldensity in cell culture manufacturing processes using radio-frequencyimpedance. Cytotechnology 50:35–48.

Cemazar J, Vrtavcnik D, Amon S, Kotnik T. 2011. Dielectrophoretic field-flow microchamber for separation of biological cells based on theirelectrical properties. IEEE Trans Nanobiosci 10.

Demierre N, Braschler T,Muller R, Renaud P. 2008. Focusing and continuousseparation of cells in amicrofluidic device using lateral dielectrophoresis.Sens Actuators B 132:388–396.

Ducommun P, Kadouri A, von Stockar U, Marison IW. 2002. On-linedetermination of animal cell concentration in two industrial high-density culture processes by dielectric spectroscopy. Biotechnol Bioeng77:316–323.

Duncan L, Shelmerdine H, Hughes MP, Coley HM, Hübner Y, Labeed FH.2008. Dielectrophoretic analysis of changes in cytoplasmic ion levels dueto ion channel blocker action reveals underlying differences betweendrug-sensitive and multidrug-resistant leukaemic cells. Phys Med Biol53:N1–N7.

Flanagan LA, Lu J, Wang L, Marchenko S, Jeon NL, Lee AP, Monuki ES. 2008.Unique dielectric properties distinguish stem cells and their differenti-ated progeny. Stem Cells 26:656–665.

Foster KR, Schwan HP. 1989. Dielectric properties of tissues and biologicalmaterials: A critical review. Crit Rev Biomed Eng 17:25–104.

Gagnon Z. 2011. Cellular dielectrophoresis: Applications to the characteri-zation, manipulation, separation and patterning of cells. Electrophoresis32:2466–2487.

Geske FJ, LiebermanR, Strange R, Gerschenson LE. 2001. Early stages of p53-induced apoptosis are reversible. Published online: 22 February 2001.doi:10.1038/sj.cdd.44007868 http://www.nature.com.proxy1.lib.umani-toba.ca/cdd/journal/v8/n2/full/4400786a.html.

Harris CM, Todd RW, Bungard SJ, Lovitt RW, Morris JG, Kell DB. 1987.Dielectric permittivity of microbial suspensions at radio frequencies: Anovelmethod for the real-time estimation ofmicrobial biomass. EnzymeMicrob Biotechnol 9:181–1186.

Huang Y, Holzel R, Pethig R, Wang X-B. 1992. Differences in the ACelectrodynamics of viable and non-viable yeast cells determined throughcombined dielectrophoresis and electrorotation studies. Phys Med Biol37:1499–1517.

Hughes JFM, Bortner CD, Purdy GD, Cidlowski JA. 1997. Intracellular Kþsuppresses the activation of apoptosis in lymphocytes. J Biol Chem272:30576.

Jaitovich AA, Bertorello AM. 2006. Naþ, Kþ-ATPase: An indispensable ionpumping-signaling mechanism across mammalian cell membranes.Semin Nephrol 26:386–3392.

De JesusM,WurmFM. 2011.Manufacturing recombinant proteins in kg-tonquantities using animal cells in bioreactors. Eur J Pharm Biopharm78:184–188.

Jones TB. 1995. Electromechanics of particles. Cambridge University Press.Kerr JFR, Wyllie AH, Currie AR. 1972. Apoptosis: A basic biological

phenomenon with wide ranging implications in tissue kinetics. Br JCancer 26:239–257.

Kim NS, Lee GM. 2002. Response of recombinant Chinese hamster ovarycells to hyperosmotic pressure: Effect of Bcl-2 overexpression. JBiotechnol 95:237–248.

Knabben I, Regestein L, Grumbach C, Steinbusch S, Kunze G, Büchs J. 2010.Online determination of viable biomass up to very high cell densities inArxula adeninivorans fermentations using an impedance signal. JBiotechnol 149:60–66.

Koopman G, Reutelingsperger CP, Kuijten GA, Keehnen RM, Pals ST, vanOers MH. 1994. Annexin V for flow cytometric detection ofphosphatidylserine expression on B cells undergoing apoptosis. Blood84:1415–1420.

Labeed FH, Coley HM, Hughes MP. 2006. Differences in the biophysicalproperties of membrane and cytoplasm of apoptotic cells revealed usingdielectrophoresis. Biochim Biophys Acta 1760:922–929.

Lim Y,Wong NSC, Lee YY, Ku SCY,Wong DCF, YapMGS. 2010. Engineeringmammalian cells in bioprocessing—Current achievements and futureperspectives. Biotechnol Appl Biochem 55:175–189.

Limoli CL, HartmannA, Shephard L, Yang C, BoothmanDA, Bartholomew J,Morgan WF. 1998. Apoptosis, reproductive failure, and oxidative stressin Chinese hamster ovary cells with compromised genomic integrity.Cancer Res 58:3712–3718.

Lucas BK, Giere LM, DeMarco RA, Shen A, Chisholm V, Crowley CW. 1996.High-level production of recombinant proteins in CHO cells using adicistronic DHFR intron expression vector. Nucleic Acids Res 24:1774–1779.

Martin SJ, Reutelingsperger CP,McGahonAJ, Rader JA, van Schie RC, LaFaceDM, Green DR. 1995. Early redistribution of plasma membranephosphatidylserine is a general feature of apoptosis regardless of theinitiating stimulus: Inhibition by overexpression of Bcl-2 and Abl. J ExpMed 182:1545–1556.

Mastrangelo AJ, Hardwick JM, Zou S, Betenbaugh MJ. 2000. Part II.Overexpression of bcl-2 family members enhances survival ofmammalian cells in response to various culture insults. BiotechnolBioeng 67:555–564.



http://www.nature.com.proxy1.lib.umanitoba.ca/cdd/journal/v8/n2/full/4400786a.html

http://www.nature.com.proxy1.lib.umanitoba.ca/cdd/journal/v8/n2/full/4400786a.html

Maxwell JC. 1881. A treatise on electricity and magnetism second. Oxford:Clarendon Press Vol. I. 2 vols.

Mimura Y, Lund J, Church S, Dong S, Li J, Goodall M, Jefferis R. 2001.Butyrate increases production of human chimeric IgG in CHO-K1 cellswhilst maintaining function and glycoform profile. J Immunol Methods247:205–216.

Muñoz-Pinedo C, Ruiz-Ruiz C, Ruiz de Almodóvars C, Palacios C, López-Rivas A. 2003. Inhibition of glucose metabolism sensitizes tumor cells todeath receptor-triggered apoptosis through enhancement of death-inducing signaling complex formation and apical procase-8 processing.J Biol Chem 278:12759–12768.

Nikolic-Jaric M, Cabel T, Salimi E, Bhide A, Braasch K, ButlerM, Bridges GE,ThomsonDJ. 2013. Differential electronic detector tomonitor apoptosisusing dielectrophoresis-induced translation of flowing cells (DEPcytometry). Biomicrofluidics 7:024101.

Opel CF, Li J, Amanullah A. 2010. Quantitative modeling of viable celldensity, cell size, intracellular conductivity, and membrane capacitancein batch and fed-batch CHO processes using dielectric spectroscopy.Biotechnol Prog 1187–1199.

PanayiotidisMI, Bortner CD, Cidlowski JA. 2006. On themechanism of ionicregulation of apoptosis: Would the Naþ/Kþ-ATPase please stand up?Acta Physiol 187:215.

Pethig R, Menachery A, Pells S, Sousa PD. 2010. Dielectrophoresis: Areview of applications for stem cell research. J Biomed Biotechnol2010:182581.

Polevaya Y, Ermolina I, Schlesinger M, Ginzburg B-Z, Feldman Y. 1999.Time domain dielectric spectroscopy study of human cells: II. Normaland malignant white blood cells. Biochim Biophys Acta 1419:257–271.

Rao RV, Niazi K, Mollahan P, Mao X, Crippen D, Poksay KS, Chen S,Bredesen DE. 2006. Coupling endoplasmic reticulum stress to the cell-death program: A novel HSP90-independent role for the smallchaperone protein p23. Cell Death Differ 13:415–425.

Schwan HP., 1957. Electrical properties of tissue and cell suspensions. In:Advances in biological and medical physics. New York Academic PressInc. Vol. V, pp. 147–209.

Simon L, Karim MN. 2002. Control of starvation-induced apoptosis inChinese hamster ovary cell cultures. Biotechnol Bioeng 78:645–657.

Strober W. 2001. Trypan blue exclusion test of cell viability. Curr ProtocImmunol 21:A.3B.1–A.3B.2.

Tey BT, Singh RP, Piredda L, Piacentini M, Al-Rubeai M. 2000. Influence ofBcl-2 on cell death during the cultivation of a Chinese hamster ovary cellline expressing a chimeric antibody. Biotechnol Bioeng 68:31–43.

Tibayrenc P, Preziosi-Belloy L, Ghommidh C. 2011. On-line monitoring ofdielectrical properties of yeast cells during a stress-model alcoholicfermentation. Process Biochem 46:193–201.

Vermes I, Haanen C, Reutelingsperger C. 2000. Flow cytometry of apoptoticcell death. J Immunol Methods 243:167–190.

Walsh G. 2010. Biopharmaceutical benchmarks 2010. Nat Biotechnol28:917–924.

Winchester B. 2005. Lysosomal metabolism of glycoproteins. Glycobiology15:1R–15R.

Wlodkowic D, TelfordW, Skommer J, Darzynkiewicz Z., 2011. Apoptosis andBeyond: Cytometry in studies of programmed cell death. Meth Cell Biol103:55–98.

Wurm FM. 2004. Production of recombinant protein therapeutics incultivated mammalian cells. Nat Biotechnol 22:1393–1398.

Yun CY, Liu S, Lim SF, Wang T, Chung BYF, Teo JJ, Chuan KH, Soon ASC,Goh KS, Song Z. 2007. Specific inhibition of caspase-8 and -9 in CHOcells enhances cell viability in batch and fed-batch cultures. Metab Eng9:406–418.

Supporting Information

Additional supporting information may be found in theonline version of this article at the publisher’s web-site.


the changing dielectric properties of cho cells can be...

Documents