Systematic dielectrophoretic analysis of the Ara-C-induced NB4 cell apoptosis combined with gene expression profiling

Abstract

Dielectrophoresis (DEP) can be used to noninvasively measure the dielectric state of the cell, and this data can be used to monitor cell health or apoptosis. In this study, we followed events associated with cytosine arabinoside (Ara-C)-induced apoptosis in NB4 cells using DEP analysis. Our data showed that the membrane capacitance of NB4 cells decreases from 9.42 to 7.63 mF/m² in the first 2 hours following treatment with Ara-C, and that this decreased capacitance persists for >12 hours. Additionally, cytoplasmic conductivity decreases from 0.217 to 0.190 S/m within 2 hours of Ara-C treatment; this level is maintained for a short period of time before decreasing. We also investigated these events molecularly at the level of gene expression using microarray analysis and showed that the expression of genes related to membrane capacitance and cytoplasmic conductivity change dramatically as early as 2 hours post-Ara-C treatment, and further demonstrated a temporal relationship between the dielectric properties and key events in apoptosis. This study, integrating physical electrical properties of the cell membrane and cytoplasm with those of conductivity-related gene networks, provides new insights into the molecular mechanisms underlying the initiation of apoptosis, establishing a systematic foundation for DEP application in follow-up drug screening and development of medicines for treating leukemia.

Introduction

Approximately 1200 new cases of acute promyelocytic leukemia (APL) are diagnosed in the United States each year (http://ghr.nlm.nih.gov/). The most widely-used anti-metabolite used to induce remission in children and adults with acute leukemia¹ and for APL consolidation therapy to reduce the relapse rate² is cytosine arabinoside (Ara-C). Ara-C is a potent killer of dividing cells by inducing cellular apoptosis mechanisms.

Apoptosis is an essential cell process for the ordered destruction of unwanted or damaged cells and can be defined as a gene-directed cellular self-destruction mechanism. Dysfunction of this mechanism is linked to the pathogenesis of cancer and other diseases³ whereby reduced apoptosis prolongs the persistence of transformed cancer cells. Thus, in vitro technologies for sensitive and rapid assessment of apoptotic events may be useful for understanding cellular events involving the apoptosis mechanism and for aiding disease treatment and anti-cancer drug development.

Many features can be used to distinguish normal cells from cells undergoing apoptosis, such as decreased mitochondrial transmembrane potential,⁴ externalization of phosphatidylserine (PS),⁵ changes in morphological features^6,7 and changes in intracellular ion concentrations.⁸ Hence, various methods have been developed to detect apoptosis. The most commonly used method is in vitro measurement of fluorescence probes designed for specific labeling of relevant molecules in apoptotic cells, such as Annexin V and JC-1 assays.^9–11 1 However, these approaches are typically invasive, resulting in cell death.

It has been long recognized that dielectrophoresis (DEP) can be used to determine the dielectric properties of a cell. DEP is noninvasive, does not require markers or labels, and has been developed for a broad range of applications. For instance, DEP can be used for cell characterization,¹² cell isolation,¹³ cell separation,¹⁴ drug resistance monitoring,¹⁵ drug assessment,¹⁶ in vitro fertilization,¹⁷ and stem cell research,^18,19 as well as for analyzing cells in different physiological states, particularly apoptosis.^20–24 However, the exact molecular mechanisms that occur during apoptosis remain unclear, as are the events associated with Ara-C-induced apoptosis.

In this study, Ara-C-induced NB4 cell apoptosis was monitored using DEP analysis and validated using standard Annexin V and JC-1 assays. A 22 K human genome oligo array was also used to examine expression levels of genes associated with membrane capacitance and cytoplasmic conductivity in parallel with DEP measurements to comprehensively investigate molecular genetic events that occur during apoptosis.

Material and methods

Cell culture and induction of apoptosis

The NB4 cell line was obtained from Peking Union Medical College. The cells were maintained in RPMI 1640 media (Gibco, Grand Island NY, USA) supplemented with 10% (v/v) fetal bovine serum (FBS; PAA, Pasching, Austria), and 1% (v/v) penicillin/streptomycin (Invitrogen, Carlsbad CA, USA). The cells were grown at 37°C in a 5% CO₂ atmosphere. Ara-C was purchased from Sigma-Aldrich (St Louis, MO, USA). To induce apoptosis, NB4 cells (at a density of 3 × 10⁵ cells/mL) were treated with Ara-C at a final concentration of 1 × 10⁻⁵ M. Concentration was determined using CCK-8 assays (see Supplementary material). The incubation period included four time points: 2, 4, 6, and 12 hours.

Detection of apoptosis using Annexin V and JC-1 assays

Apoptosis was analyzed using flow cytometric analysis (FCA). Cells were cultured in 6-well plates, treated with 1 × 10⁻⁵ M Ara-C, collected at different time points, and centrifuged at 190 × g for 5 minutes. A rapid Annexin V-FITC apoptosis detection kit (KeyGEN Biotech, Nanjing, China) was used according to the manufacturer’s instructions. Analysis was performed on a BD FACSAria™ II system (BD Biosciences, San Jose, CA, USA). Cells incubated without Ara-C were analyzed in parallel as controls.

JC-1 was used to dye NB4 cells to visually detect early apoptosis. Similarly to the procedure for flow cytometric analysis, cells were collected and incubated in the dark in 0.5 mL incubation buffer (KeyGEN Biotech) with 1 μL JC-1 (2.5 mg/mL) at 37°C for 20 minutes. After incubation, the cells were washed twice with 1 × PBS and analyzed using fluorescence microscopy (DM-IRB; Leica, Wetzlar, Germany). To reduce the presence of dye particulates prior to incubation, the JC-1 solution was sonicated for 5 minutes, followed by centrifugation (1 minute at 9300 × g). The supernatant was used for dye loading.

Dielectrophoretic system setup and crossover frequency determination

The principal components of the DEP system are described in Figure 1. The DEP chip was fabricated by patterning nonclosed ring gold electrodes onto the surface of a clean glass wafer using standard photolithography. As shown in Figure 1A, the width of the electrodes on the glass chip was 20 μm and the inner diameter of the ring was 200 μm. A 200-μL Eppendorf tube was cut and assembled on the chip to construct the DEP measurement chamber (Figure 1B).

Figure 1.

Principal components of the DEP assay system. (A) Nonclosed ring gold electrodes, with width of 20 μm and inner diameter of 200 μm. (B) The assembled chip with a reservoir modified from a 200-μL Eppendorf tube. (C) Signal generator. C1: HP33120A, C2: SMB100A. (D) CCD camera coupled to a fluorescence microscope.

Abbreviation: DEP, dielectrophoresis.

The sine signal applied to the chip was generated by a signal generator (HP33120A; Hewlett-Packard Santa Clara, CA, USA) in the range of 10–15 MHz or a signal generator (SMB100A; Rohde and Schwarz, Munich, Germany) in the range of 50–500 MHz at an applied voltage of 3 V (peak-to-peak). Measurements were observed through a CCD camera (DP-71; Olympus, Tokyo, Japan) coupled to a fluorescence microscope (DM-IRB; Leica).

Prior to DEP measurements, the cells were centrifuged at 190 × g for 5 min. The pellets were washed and resuspended in a prepared isotonic medium containing 8.5% (w/v) sucrose and 0.3% (w/v) dextrose buffer for which conductivity had been adjusted to 32.8 mS/m with RPMI 1640 media and a conductivity meter (DDSJ-308A; SPSIC Ltd Shanghai, China). Subsequently, the cells were added to the chip chamber using a micropipette, and the motion of the cells toward or away from the electrode edges due to the applied frequency was observed.

DEP measurements are made by determining crossover frequencies, at which most of the cells in the chamber experience a zero DEP-induced force and exhibit no movement (ie, corresponding to the transition between negative and positive DEP). The crossover frequency for ≥20 cells was measured over a time period of ≤20 minutes for each experiment. The frequency obtained was further used to determine the capacitance of the plasma membrane²³ or cytoplasmic conductivity.²⁵

Gene expression profiling and real-time quantitative polymerase chain reaction (PCR)

The 22 K human genomic oligo array (CapitalBio Corp, Beijing, China) contains 21,329 5′-amino-modified 70-mer probes of the Human Genome OligoSet (Version 2.1; Operon, Huntsville, AL, USA). Total RNA was isolated using the Trizol method (Invitrogen). Reverse transcription was performed using M-MLV (Takara Chemicals, Shiga, Japan). Microarray experiments were performed as described previously.²⁶ After hybridization, microarrays were scanned using a LuxScan™ 10 K/A confocal scanner (CapitalBio), and data from the obtained images were extracted using LuxScan 3.0 software (CapitalBio). Raw data were normalized using the space- and intensity-dependent LOWESS program.²⁷ For each test and control sample at the 2-, 4-, 6-, and 12-hour time points, hybridization experiments were performed using the dye-swap strategy. Only genes with consistent differential expression (both above a twofold change) in both dye-swap microarrays were selected as differentially expressed genes. The description of this microarray study follows the Minimum Information About a Microarray Experiment (MIAME) guidelines.²⁸

A selected subset of differentially expressed genes was validated using real-time quantitative PCR employing an EvaGreen Real-time qPCR Core Reagent Kit (CapitalBio) and a Bio-Rad IQ™ 5 (Bio-Rad, Hercules, CA, USA). Target genes and a reference gene (glyceraldehyde 3-phosphate dehydrogenase [GAPDH]) were amplified in parallel. Primer sequences are shown in Supplementary Table S1. The results were analyzed using Bio-Rad IQ™ 5 (software version 2.1; Bio-Rad). PCR amplification products were analyzed using melting curve analysis and 1.5% agarose gel electrophoresis. The formula ${\displaystyle {\text{E}}_{\text{target}}^{\text{CP}1–\text{CP}2}/{\text{E}}_{\text{GADPH}}^{\text{CP}3–\text{CP}4}}$ was employed to calculate the relative gene expression ratio of a target gene, where E is the real-time PCR efficiency and CP is the real-time PCR crossing point.²⁹ All reactions were performed in triplicate.

Detection of morphology changes using scanning electron microscopy (SEM) and indirect immunofluorescence (IF) microscopy

After centrifugation at 190 × g for 5 minutes followed by resuspension, the cells were fixed with 2.5% glutaraldehyde at room temperature for 1 hour, washed in 1 × PBS, and exposed to 1% osmium tetroxide for another hour. The cells were dehydrated by sequential immersion in 30, 50, 70, 80, 90, and 2 × 100% ethanol, followed by a 50/50 ethanol/tert-Butanol solution, and then 100% tert-Butanol. The cells were freeze-dried (ES-2030; Hitachi, Tokyo, Japan) for 4 h, coated with a thin gold layer using an ion sputter (E-1010; Hitachi), and finally attached to the microscope supports using silver glue. The cells were imaged at 15 kV by SEM (Quanta 200; FEI, OR, USA).

For IF analysis, cells were collected washed and plated on glass coverslips. Next, the cells were fixed in 4% paraformaldehyde for 30 min and permeabilized using 0.2% Triton X-100 in 1 × PBS. After blocking with 1% (w/v) bovine serum albumin (BSA) in 1 × PBS, the cells were incubated with rabbit KIF20A polyclonal antibody (Abnova, Taipei, Taiwan) or rabbit y-tubulin polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) at 4°C overnight. Subsequently, the cells were washed three times with 1 × PBS and stained with DyLight-labeled goat anti-rabbit IgG (Jackson, West Grove, PA, USA) at room temperature for 4 hours. Nuclei were stained with 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI); Vector Labs, Burlingame, CA, USA). The cells were then examined and photographed using a laser scanning confocal microscope (LSM710; Carl Zeiss, Oberkochen, Germany). The excitation wavelength was 488 nm for DyLight and 405 nm for DAPI. Images were merged using ZEN software (Carl Zeiss).

Measurements of intracellular calcium, sodium, and potassium using flow cytometry

Intracellular calcium ion concentration was measured using Fluo4-AM (Dojindo, Kumamoto, Japan) and 20% pluronic F-127 dissolved in DMSO to make a 1 mM stock solution. Immediately before Fluo 4-AM loading, the cells were gently washed with HEPES-buffered saline (10 mM HEPES, 137 mM NaCl, 5 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 10 mM D-glucose, pH 7.35) and then incubated in Fluo 4-AM working solution (5 μM Fluo 4-AM and 0.05% pluronic F-127 dissolved in HEPES buffer) for 30 minutes at 37°C in a cell incubator. Next, the cells were washed with HEPES-buffered saline to remove extracellular Fluo 4-AM and incubated for an additional 20 minutes to allow deesterification of Fluo 4-AM into its active-dye form Fluo 4. Analysis was performed on a BD FACSCalibur™ system (BD Biosciences) at an excitation of 488 nm.

For intracellular sodium and potassium ion measurements, 2 μL of 2.5 mM SBFI-AM (Sigma-Aldrich) and PBFI-AM (Santa Cruz Biotechnology) stock solutions were individually added to 1 mL of cells at a final concentration of 5 μM 1 hour prior to conducting the experiment. Incubation was continued at 37°C in a 5% CO2 atmosphere. Immediately before FCA, propidium iodide (PI; Sigma-Aldrich) was added at a final concentration of 10 μg/mL. Approximately 1 × 10⁴ cells were analyzed by sequential excitation of the cells at 355 nm for SBFI-AM or PBFI-AM and 488 nm for PI, using a BD™ LSR II Flow Cytometer System (BD Biosciences).

Results and discussion

Detecting apoptosis with Annexin V and JC-1 assays

Apoptotic cells display significant externalization of PS, which can be validated using the Annexin V assay. As shown in Supplementary Figure S2, a series of the Annexin V assays based on flow cytometry was performed at different time points. Our results demonstrate that the proportion of Annexin V-positive cells at 2 hours posttreatment was not significantly different compared to control cells, and up to 4 hours posttreatment, the proportion of the positive cells displayed a clear difference (Figure 2A). Among the entire cell population, this proportion increased from 4.51% ± 0.44% at 2 hours to 7.19% ± 0.26% at 4 hours, 12.42% ± 0.68% at 6 hours, and 21.68% + 0.45% at 12 hours.

Figure 2.

Detection of apoptosis by Annexin V and JC-1 assays. (A) Percentage of apoptotic cells after different incubation periods with Ara-C as measured using FCA and Annexin V/PI assays. (B) Viable cells display greenish orange fluorescence and apoptotic cells display green fluorescence.

Note: Our results suggest that NB4 cell apoptosis initiated by Ara-C can be detected at 2 h post-treatment using the JC-1 assay.

Abbreviations: Ara-C, cytosine arabinoside; FCA, flow cytometric analysis.

To ensure that apoptosis initiated by Ara-C could be detected as early as possible (ie, at 2 hours posttreatment), assays based on the mitochondria membrane potential (ΔΨm) were performed. A typical early characteristic of apoptosis is permeabilization of the mitochondrial membrane followed by dissipation of ΔΨm, which is measured using the JC-1 assay¹⁰ As shown in Figure 2B, cells displaying greenish orange fluorescence were viable, while cells displaying green fluorescence were apoptotic. After 2 hours of Ara-C treatment, the proportion of apoptotic cells increased. This indicates that NB4 cell apoptosis initiated by Ara-C can be detected as early as 2 hours posttreatment using the JC-1 assay, during which time the mitochondrial permeability transition pore opens and the corresponding ΔΨm is reduced.

DEP analysis during Ara-C-induced apoptosis

DEP analysis is an assay based on measuring changes in dielectric properties (ie, membrane capacitance and cytoplasmic conductivity) that has been applied for monitoring cell apoptosis.^20–24 Currently, there are several theoretical models used to extract cell dielectric properties.^30,31 In this study, we examined the variation in cell dielectric property trends during cell apoptosis. We applied a simple model for measuring crossover frequencies in an isotonic medium with a specific conductivity (32.8 mS/m).^23,25 DEP analysis reveals two types of crossover frequencies, the lower and upper crossover frequencies, which are indicated by ƒ_x1 and ƒ_x2, respectively. ƒ_x1 provides a direct measurement of membrane capacitance, while f_x2 provides a direct measurement of cytoplasmic conductivity.²⁵ As shown in Figure 3A and B, crossover frequency of the cells at 2 h post-exposure to Ara-C displayed distinct differences with that obtained from control cells; moreover, ƒ_x1 continually increased from 96 ± 4.73 to 354 ± 6.11 kHz, while f_x2 decreased from 301 ± 7.09 to 165 ± 7.78 MHz over the 12-hour time course. Additionally, cell diameter decreased from 16.30 ± 0.35 to 13.81± 0.62 μm (Figure 3C). As shown in Table 1, membrane capacitance (C_m) decreased from 9.42 ± 0.47 to 3.03 ± 0.05 mF/m² over the 12-hour time course based on the formula of Pethig and Talary $(C_m=\sqrt{2}/2\pi r{f}_{X1}{\sigma }_s)$,²³ where r and σ_s represent the cell radius and conductivity of the surrounding solution (32.8 mS/m), respectively. The decreasing trend of membrane capacitance observed in our experiments is consistent with the observations of previous studies.^23,24 Similarly, we calculated that cytoplasmic conductivity (σ) decreased from 0.217 ± 0.005 to 0.120 ± 0.006 S/m over the 12-hour time course based on the equation $({\sigma }_{cyto}=2\sqrt{2}\pi {\epsilon }_s{f}_{X2})$,²⁵ where ɛ_s indicates the relative permittivity of the surrounding solution is a constant, about 81. This generally agrees with previous studies examining Jurkat T-cells undergoing etoposide-induced apoptosis.²³ Compared to the Annexin V and JC-1 assays, the DEP assay can be used to monitor apoptosis noninvasively without the use of added markers or labels.

Figure 3.

Monitoring of Ara-C induced apoptosis by DEP analysis. (A and B) Distinct differences in crossover frequency occur between control and the apoptotic NB4 cells as early as 2 h post-exposure to Ara-C. ƒ_x1 increased from 96 ± 4.73 to 354 ± 6.11 KHz, while ƒ_x2 decreased from 301 ± 7.09 to 165 ± 7.78 MHz over the 12-h time course. (C) Cell diameter decreased from 16.30 ± 0.35 to 13.81 ± 0.62 μm in apoptotic cells.

Note: No significant change was observed for control cells.

Abbreviations: Ara-C, cytosine arabinoside; DEP, dielectrophoresis.

Table 1.

Changes in membrane capacitance (C_m) or cytoplasmic conductivity (σ_cyto) during Ara-C-induced apoptosis

Time (hours)	Cm(mF/m2)	σcyto (S/m)
0 (control)	9.42 ± 0.47	0.217 ± 0.005
2	7.63 ± 0.25	0.190 ± 0.005
4	5.75 ± 0.21	0.188 ± 0.004
6	4.93 ± 0.20	0.160 ± 0.004
12	3.03 ± 0.05	0.120 ± 0.006

Gene expression profiling during Ara-C-induced apoptosis

To better understand the molecular mechanism underlying apoptosis using DEP monitoring, we examined mRNA changes in NB4 cells after Ara-C treatment over a 12-hour time course using a human whole genome oligo array. A total of 9, 37, 42, and 117 genes displayed altered expression levels at 2, 4, 6, and 12 hours, respectively, as shown in Supplementary Table S2. A total of 152 genes showing ≥2.0-fold change at least once for all four time points (Supplementary Table S3) were selected for gene ontology (GO) analysis and cluster analysis (Supplementary Figure S3).

GO analysis revealed that differentially expressed genes can be grouped into five main categories: apoptosis, cell division and proliferation, cell morphogenesis, ion transport and cell polarity, and others. Additionally, cluster analysis confirmed the expression level of some known apoptosis-related genes changed after Ara-C treatment, including the up-regulated pro-apoptosis genes SESN2, E2F2, DDIT4, TNF, GADD45A, and PMAIPI and the downregulated antiapoptosis genes HSPA1A, HSPA1B, and ARHGDIA (Figure 4, Supplementary Figure S4(A)). In contrast, the expression level of nearly all genes involved in cell division and proliferation was downregulated (eg, C20orf129 and IRS2) (Figure 4, Supplementary Figure S4(B)). Notably, the expression level of some genes involved in cell morphogenesis or ion transport changed as early as 2 hours post-Ara-C treatment, which may correlate with the measured alteration in membrane capacitance or alteration in cytoplasmic conductivity (Figure 4, Supplementary Figure S4(C and D)). Furthermore, we analyzed expression changes in terms of transcriptional regulation of defined gene networks and pathways and present a model of the dynamic and temporal network of pathway activities related to Ara-C-induced apoptosis (Supplementary Figure S5), which may have implications for identifying potential new effectors useful for treating APL.

Figure 4.

Relationship of molecular events to the time course of Ara-C-induced apoptosis.

Note: Cluster image of gene expression data of the four main categories: apoptosis, cell division and proliferation, cell morphogenesis, and ion transport and cell polarity.

Abbreviation: Ara-C, cytosine arabinoside.

The original data from this study are available at the Gene Expression Omnibus (see http://www.ncbi.nlm.nih.gov/geo); the accession number is GSE32992. Results of real-time quantitative PCR analysis for all selected genes were in agreement with microarray data (Supplementary Figure S4).

Systematic analysis of changes in dielectric properties

Because DEP analysis is noninvasive, nondestructive, and rapid it has been widely employed for cell apoptosis studies.^20–24 However, little is known regarding the temporal relationship between dielectric properties measured using DEP analysis and key events in apoptosis. Here, we used DEP analysis to systematically investigate potential mechanisms responsible at the gene level and at the cellular level for changes in membrane capacitance and cytoplasmic conductivity occurring during apoptosis.

As described above, membrane capacitance decreased from 9.42 ± 0.47 to 3.03 ± 0.05 mF/m² over 12 hours. Cell membrane capacitance largely depends on plasma membrane surface morphology, such as microvilli, ruffles, folds, and blebs;^32–35 cells treated with Ara-C, as well as control cells, were imaged using SEM (Figure 5A). Control cells were covered with numerous microvilli, which appeared to be homogeneous both in size (~16 μm diameter) and surface morphology. However, treated cells exhibited alterations in their plasma membrane morphology. After 2 hours of treatment with Ara-C, the number of microvilli slightly decreased, and the membrane surface became smoother. After a treatment time of ≥4 hours, cell shrinkage and bleb formation were observed. Membrane surface morphology is closely associated with the cytoskeleton³⁶ and thus can be monitored by observing changes in microtubule-related proteins during apoptosis.^37,38 As shown in Figure 5B, the average cell fluorescence intensity (FI) related to γ-tubulin from the IF assay gradually decreased over the 12-hour time course, indicating that in cells treated with Ara-C, micro-tubules were disrupted and collapsed, resulting in obvious changes to the cytoskeleton as observed by others.³⁷ Changes in surface morphology involving the cytoskeleton effectively decreased the surface area of the cell membrane and, hence, decreased cell membrane capacitance.

Figure 5.

Changes in morphological features of cells over the 12-h time course of Ara-C induced apoptosis. (A) SEM images of NB4 cells after treatment for 0 (control), 2, 4, 6, and 12 h with Ara-C. The cells displayed progressive changes in membrane morphology as apoptosis developed. (B and C) Merged confocal scanning laser microscopy images represent the changes in microtubule-related proteins over the 12-h time course: (B) γ-tubulin and (C) KIF20A.

Note: Cell nuclei are stained with DAPI (blue).

Abbreviations: Ara-C, cytosine arabinoside, SEM, scanning electron microscopy; DAPI, 4′,6-diamidino-2-phenylindole dihydrochloride.

We examined the expression level of microtubule-related genes, such as kinesin family member 20A (KIF20A)³⁹ and centromere-associated protein E (CENPE),^40,41 and found that these genes were downregulated over the 12-hour time course. More precisely, expression of KIF20A and CENPE decreased to a minimum at 4 hours posttreatment and then remained low for the remainder of the time course (Supplementary Figure S4(C)). Accordingly, the decline in average cell fluorescence intensity related to KIF20A in the IF assay reflects the decreasing KIF20A protein quantity and activity (Figure 5C), which was similar to γ-tubulin. Because of the close association between membrane surface morphology and KIF20A, a cytoskeletal component,³⁶ the decline in KIF20A expression may induce structural changes to the cell membrane (Figure 5A), further decreasing cell membrane capacitance.^33,34 Subsequently, an increase in ƒ_x1 was observed as a consequence of these changes.

Additionally, cytoplasmic conductivity decreased from 0.217 ± 0.005 to 0.120 ± 0.006 S/m. σ_cyto is another important cell dielectric property which is, to a large extent, determined by ion mobility and concentration in the cytoplasm.⁴² Our gene expression profiling results demonstrated that expression of ion transport-related genes, such as P2X purinoceptor 4 (P2RX4; associated with calcium and sodium transport)^43,44 and potassium channel tetramerisation domain containing 9 (KCTD9; associated with potassium transport),^45,46 was up-regulated over the 12-hour time course (Supplementary Figure S4(D)). Moreover, changes in intracellular calcium, sodium, and potassium concentration were measured using fluorescence probe-based flow cytometry, and all changes in cytoplasmic conductivity over the 12-hour time course were estimated using the equation $(k=F\sum |Z_i|C_i{U}_i)$ of Bard and Faulkner.⁴⁷ The experimental data revealed an increase in intracellular calcium and sodium, as well as a loss of intracellular potassium, during apoptosis (Supplementary Figure S6(A–C)). Figure 6 shows the relative dynamic and temporal changes in intracellular ions and cytoplasmic conductivity over the 12-hour time course. Intracellular calcium and sodium increased to >1.5 and 1.2 times their normal levels, respectively, while intracellular potassium level was reduced to <0.80 times its normal level over the 12-hour time course (Figure 6A). A decrease in intracellular potassium ultimately induces a loss of cytoplasmic conductivity because it is a highly abundant intracellular ion,⁴⁸ even when intracellular calcium and sodium levels increase. Estimated relative changes of average cytoplasmic conductivity based on ion levels decreased significantly at 2 hours post-Ara-C treatment and changed minimally thereafter. T DEP analysis showed nearly the same results over the 6-hour time course. However, the difference became significant at the 12-hour time point (Figure 6B). One possible reason for this difference may be that the ions contributing to cytoplasmic conductivity were limited. Based on gene expression profiling, we found that the expression level of chloride intracellular channel 4 (CLIC4; associated with chloride transport)^49,50 was significantly up-regulated at the 12-hour time point (Supplementary Table S3), indicating that an apparent change of intracellular chloride ions may have occurred, further affecting overall cytoplasmic conductivity. As a label-free method for measuring overall cytoplasmic conductivity, DEP analysis may be a better choice as a simple and quick monitoring method.

Figure 6.

Relative dynamic and temporal changes in intracellular ion levels and cytoplasmic conductivity over the 12-h time course. (A) Fold changes in intracellular calcium, sodium, and potassium from the 0- to 12-hour time points during Ara-C-induced apoptosis. The data represent averages of three independent experiments. (B) Estimated changes in average cytoplasmic conductivity based on changes in intracellular ions levels (blue) or DEP analysis (red) over the 12-h time course.

Abbreviations: Ara-C, cytosine arabinoside; DEP, dielectrophoresis.

Conclusion

In this study, we used DEP analysis to investigate the mechanisms responsible for apoptosis from the gene and cell levels by performing a series of biological assays (gene expression profiling, SEM, IF, and FCA). At the gene level, our studies indicate that expression levels of KIF20A and CENPE, which are related to membrane capacitance, are down-regulated, while expression levels of P2RX4 and KCTD9, which are related to cytoplasmic conductivity, are up-regulated as early as 2 hours post-Ara-C treatment. Accordingly, at the cellular level, the FI related to γ-tubulin and KIF20A gradually decreased, and obvious structural changes to the cell membrane were observed, leading to the loss of membrane capacitance by decreasing the effective surface area of the cell membrane. Additionally, a decrease in intracellular potassium concentration induces a loss of cytoplasmic conductivity, even though the intracellular calcium and sodium levels increase. All of these changes in cell membrane capacitance and cytoplasmic conductivity make apoptotic cells amenable for characterization using DEP analysis, which may be used to help physicians detect apoptosis earlier. As this technology develops, individually tailored and more personalized patient treatment will be possible.

Supplementary material

Methods

Determining Ara-C concentration The appropriate working concentration of Ara-C was evaluated using Cell Counting Kit-8 (CCK-8; Dojindo, Kumamoto, Japan) assays. Cells were plated at a density of 1 × 10⁵ cells/mL in a 96-well plate. Incubation times and concentrations of Ara-C are shown with the results. After incubation, cell viability was determined using CCK-8 according to the manufacturer’s protocol. Absorbance at 450 and 630 nm was read using a Microplate Reader (Bio-Rad). The experiment was repeated in triplicate.

Results and discussion

As shown in Supplementary Figure S1, several different concentrations were used to test the effect of Ara-C on cell viability at various time points. IC₅₀ (defined as the concentration of drug that reduces cell viability by 50% relative to untreated controls) values were calculated to be approximately 10⁻⁴ or 10⁻⁵ M for 12 h. Combined with the results of a previous report,⁵¹ a concentration of 10⁻⁵ M Ara-C was selected to induce cell apoptosis.

Figure S1

Inhibition of growth of NB4 cells by Ara-C.

Note: Relative viabilities were determined using CCK-8 assays as described in the experimental section and the results were compared with untreated (control) cells.

Abbreviations: Ara-C, cytosine arabinoside; CCK-8, Cell Counting Kit-8.

Figure S2

Example of FCA of a single cell apoptosis experiment.

Note: Five time points are shown for FCA, 0, 2, 4, 6, and 12 hours after addition of 10⁻⁵ M Ara-C; PI⁺/FITC⁺ indicates deeply apoptotic or dead cells, while PI⁻/FITC⁺ indicates apoptotic cells.

Abbreviations: Ara-C, cytosine arabinoside; FCA, flow cytometric analysis; h, hours.

Figure S3

Gene expression profiling using microarray. (A) Gene ontology image of differentially expressed genes. (B) Cluster image of gene expression profiles to Ara-C-induced apoptosis from 0- to 12-hour time points.

Abbreviations: Ara-C, cytosine arabinoside; h, hours.

Figure S4

Real-time quantitative PCR analysis of the expression of selected genes. The results of real-time quantitative PCR analysis were similar to those microarray gene expression data. (A) Apoptosis-related genes. (B) Cell division- and proliferation-related genes. (C) Microtubule-related genes. (D) Ion transport-related genes.

Abbreviations: PCR, polymerase chain reaction.

Figure S5

Systematic transcription regulation pathway map of the apoptosis time course.

Notes: Solid dark lines represent previously known pathways; dashed dark lines represent undefined pathways. Red rectangles represent induced genes with fold changes ≥ 1.5; green rectangles represent repressed genes with fold changes ≥ 0.67. Blue ovals represent hypothetical activated TFs. Further TF ELISA experiments are needed to determine whether p53 or NF-κB activation is involved in Ara-C-induced apoptosis of NB4 cells. All elements are approximately arranged relative to the time axis.

Abbreviations: Ara-C, cytosine arabinoside; NF-κB, nuclear factor κB; TF, transcription factor.

Figure S6

Changes in intracellular calcium, sodium, and potassium over the 12-h time course measured using FCA. (A) An increase in Fluo 4-AM (Ca²⁺) fluorescence from control cells indicates an increase in intracellular calcium. (B) An increase in SBFI-AM (Na⁺) fluorescence from control cells indicates an increase in intracellular sodium. (C) A decrease in PBFI-AM (K⁺) fluorescence from control cells indicates a loss of intracellular potassium.

Notes: The data are shown as histograms which are representative of a single experiment.

Abbreviations: FCA, flow cytometric analysis h, hours.

Table S1

Sequences of primer pairs used for real-time quantitative PCR

Gene	Forward primer (5′–3′)	Reverse primer (5′–3′)
SESN2	AGATGGAGAGCCGCTTTGAGCT	CCGAGTGAAGTCCTCATATCCG
E2F2	AGCTGGAACCGAGAGAACATG	ACACGACCAGGCGAAACC
TNF	GGCGTGGAGCTGAGAGATAAC	GGTGTGGGTGAGGAGCACAT
HSPA1A	ACCTTCGACGTGTCCATCCTGA	TCCTCCACGAAGTGGTTCACCA
HSPA1B	ACCTTCGACGTGTCCATCCTGA	TCCTCCACGAAGTGGTTCACCA
C20orf129	GCGTGGCAACAGGCTCCTACA	GGCATACAGGATTCGGAACTCC
IRS2	CCTGCCCCCTGCCAACACCT	TGTGACATCCTGGTGATAAAGCC
DDIT4	AGGAAGCTCATTGAGTTGTG	GGTACATGCTACACACACAT
KIF20A	GTACCAACCAGGAAAATCAG	TGTCTGAGTATTGCATCCTG
CENPE	GGAGAAAGATGACCTACAGAGGC	AGTTCCTCTTCAGTTTCCAGGTG
P2RX4	GTGGCGGATTATGTGATACCAGC	CACACAGTGGTCGCATCTGGAA
KCTD9	GCCGCTGTAATCTTGCACATGC	CAGTTTCAGGGATGCTCCTTCTG
GAPDH	CGCTCTCTGCTCCTCCTGTT	CCATGGTGTCTGAGCGATGT

Abbreviation: PCR, polymerase chain reaction.

Table S2

Number of differentially expressed genes

Time (hours)	Up	Down	Total
2	5	4	9
4	15	22	37
6	24	18	42
12	95	22	117

Table S3

Total of 152 genes with ≥2.0-fold change at least once for all four time points

Oligo_id	Gene	Ratio
		2 hours	4 hours	6 hours	12 hours
H200015775	TNF	2.2768	1.6131	1.8894	2.1108
H200019847	DDIT3	1.1387	2.8434	3.7666	4.1329
H200011743	DDIT4	1.865	5.208	5.8026	5.2344
H200001568	SESN2	1.1782	2.5341	2.9756	2.1333
H200008354	TXNIP	1.1029	2.4659	2.6146	3.029
H200015433	E2F2	1.3478	2.0973	2.3398	3.3365
H200000027	PMAIP1	1.2504	1.9617	3.1681	3.0868
H200006759	GADD45A	1.11	1.7016	2.1676	2.1329
H200012407	CASP8AP2	1.4683	1.7928	2.1279	2.1613
H200003690	TRIB3	1.2974	1.9831	3.8015	3.8056
H200006040	CHAF1B	1.2458	1.6792	2.2018	1.8877
H200005977	PRNP	1.2409	1.2069	1.4649	2.3403
H200006833	NFKBIA	1.0624	1.1148	1.3696	2.2215
H200006782	IRF1	1.2729	1.2868	1.6145	2.0059
H200000124	ATF3	1.1052	1.6494	1.559	2.7846
H200011520	NAB1	1.3486	1.4977	1.6588	2.0771
H200010637	E2F8	1.1142	1.5773	1.8893	2.0674
H200004695	REL		1.521	1.9931	2.2944
H200015131	TNFAIP3		1.7875	3.1083	10.82
H200019156	EGR1		5.9961
H200003705	RASSF1			1.2268	2.3082
H200003433	MOAP1				2.2819
H200006172	ASNS				2.1505
H200012696	ZNF624				3.6537
H200015127	PRKCE				2.055
H200008824	KLF6				3.2992
H200006354	BTG1	0.8407	1.1771	2.1741	3.1005
H200006664	NFIL3	0.8679	1.2632	1.9841	3.1347
H200005485	PAK1	0.9977	1.1359	1.4839	2.0311
H200008426	DUSP6	0.6807	0.7826	1.0939	2.3816
H200001719	HSPA1A	0.4533	0.8086	0.6038	0.4123
H200017080	HSPA1B	0.4746	0.8206	0.5659	0.3964
H200005758	EIF4EBP1	0.5687	0.3518	0.3128	0.4273
H200007392	ARHGDIA	0.9244	0.8883	0.7877	0.4844
H200001819	CCNE1	1.4023	2.2679	2.2377	1.8367
H200006902	TIEG/KLF10	1.3368	1.8159	1.7091	2.5303
H200000106	ADM	1.0576			2.0229
H200000570	CXCL10				10.3587
H200013414	IRS2	0.4683	0.7933	1.121	1.0842
H200005718	C20orf129/FAM83D	0.491	0.4594	0.4546	0.5191
H200001767	TPX2	0.6455	0.4921	0.5083	0.6906
H200006436	DLG7/DLGAP5	0.765	0.4849	0.5174	0.445
H200010905	BUB1	0.6302	0.448	0.4873	0.5465
H200012276	ASPM	0.7851	0.4425	0.5062	0.6396
H200003347	CCNB1	0.5973	0.425	0.4063	0.4447
H200005304	KNTC2/NDC80	0.716	0.398	0.514	0.8752
H200006423	PLK1	0.6491	0.35		0.334
H200004906	CDCA8	0.7698	0.563	0.4997	0.6453
H200005764	ZNF339/OVOL2	0.7392	0.6969	0.4721	0.6271
H200006136	CENPE	0.6114	0.3547	0.4507	0.5201
H200005853	KIF20A	0.5375	0.3259	0.31	0.411
H200012386	ECT2			0.4648
H200005775	KCTD9	1.7487		2.7477	4.1513
H200018987	P2RX4		1.1764	1.7371	2.8761
H200003474	CLIC4	1.181	1.1757	1.6314	2.1938
H200005870	CCL3			2.417	3.5841
H200006175	CCL4			3.4727	3.9945
H200003915	TNFAIP6				11.9615
H200020490		2.1693	2.0409	1.4841	1.6363
H200006467	HMGCS1	2.1238	1.5023	1.1664	1.0935
H200005214	INSIG1	2.8237	1.373	1.1014	1.0765
H200004003	CCNE2	1.6662	2.6628	3.8329	3.4481
H200002187	TIPARP	1.1651	2.0555	2.1295	2.3472
H200017137	CCL3L1	1.3932	1.2863	2.057	5.6597
H200006526	ATP2B1	1.141	1.0957	1.6069	2.9833
H200002418	FOXO3A	1.1284	1.1058	1.5603	2.667
H200001281	HECA	1.0933	1.1625	1.6718	2.8152
H200005834		1.105	1.177	1.2049	2.2256
H200004109	RSN	1.0981	1.2093	1.6478	2.2426
H200002137	KIAA1147	1.1165	1.2148	1.4234	2.3482
H200002754	N4BP2	1.3414	1.2147	1.5133	2.1562
H200014202	RGS16	1.0903	1.2847	1.6735	3.4525
H200011487	CCNB1IP1	1.1918	1.2945	1.5251	2.2965
H200012358	MAP1LC3B	1.0403	1.3294	1.9677	2.3889
H200009957		1.1546	1.4168	1.7351	2.4842
H200011669	C6orf48	1.1391	1.5785	1.4732	2.7148
H200006727	MMD2	1.764	1.6883	1.8576	2.657
H200006646	CD83	1.4052	1.7151	1.8771	2.8336
H200002948	ATP2B1	1.1653	0.9668	1.2938	2.2311
H200007999	GBP7	1.1948	0.9544	1.0879	2.2038
H200007904	PTPRC	1.1887	0.8226	0.9554	2.1509
H200006209	C1orf56/MLLT11	1.2761	0.93	0.9883	2.2463
H200017557	OPN3	1.3918	1.1323	0.7939	0.4675
H200009444		1.0565	1.0179	0.9952	0.4823
H200006408	MAT2A	1.1349	1.0642	0.6771	0.4931
H200000511	TBXAS1	1.0995	0.9085	0.7469	0.4983
H200006231	KIAA0182	1.1452	0.7109	0.6425	0.4688
H200012473	PCF11	1.1002	0.8868	0.6982	0.4149
H200006260		1.1805	0.9238	0.5678	0.3931
H200002834	FAM111A	0.7931	1.1888	1.5798	2.7231
H200008575	MEF2A	0.9956	1.0874	1.2269	2.1852
H200001006	DHRS13	0.8267	1.4287	1.8371	2.1821
H200017445	SUCNR1	0.8517	0.8042	1.2275	2.6318
H200003843	SAT1	0.9061	0.9701	1.1082	2.316
H200004069	PTPRE	0.8568	0.9531	1.098	2.1504
H200006587	RGS2	0.6477	0.8868	1.4771	2.6675
H200008209	NAV3	0.6571	0.4408	1.253	3.3403
H200007845		0.9676	0.4938	0.6507	1.2268
H200021128	TCTN1	0.848	0.3728	0.4892	1.2922
H200001569	RSAD1	0.9789	0.9484	0.8524	0.4888
H200017562	C1GALT1C1	0.9565	0.7658	0.6933	0.4991
H200002814	C9orf140	0.9387	0.7971	0.7279	0.4626
H200002245	C22orf9	0.8398	0.8835	0.908	0.4672
H200020390	FAM46A	0.8073	0.7061	0.6584	0.4897
H200003471	YWHAG	0.852	0.52	0.4867	0.5355
H200000419	CENPA	0.7722	0.542	0.4533	0.6846
H200001941	FAM46A	0.7428	0.6849	0.6931	0.4513
H200016105		0.7293	0.6981	0.4905	0.792
H200008360	DHRS9	0.61	0.4867	0.6395	0.7368
H200004769	HIST1H4C	0.5819	0.4728	0.6094	0.6927
H200019346		0.6052	0.4432	0.4016	0.718
H200018740		0.5938	0.4321	0.5277	0.7306
H200001067	EVI2B	0.6304	0.4167	0.6691	0.7823
H200016261	STK6/AURKA	0.6082	0.4028	0.441	0.6585
H200013081	DEPDC1	0.5525	0.3815	0.4393
H200020123		1.0639	1.2445		2.0239
H200001749	CASP7	0.9548	1.4103		2.544
H200007958	CCDC134	0.694	0.7555		2.0645
H200018675		0.8948	0.8359		0.4542
H200015636		2.0966	1.5265
H200000156			5.1151	25.8726	126.9222
H200001516	PELI1		1.4342	1.7191	2.1665
H200011363			0.6085	0.4937
H200000205	HIST2H2AA3		0.4501	0.4727
H200015535	FICD		2.263		2.2602
H200016527	CXorf34/TRMT2B		2.0672		2.1392
H200005739	NEFM		1.4999		2.8711
H200001150	ARRDC4		2.3531		1.9712
H200016414			2.0232
H200017236	LIMK2			1.4211	2.0302
H200004185	CDKN2AIP			1.8434	2.2906
H200003339	AVPI1			1.8719	2.0933
H200019676	LRP11			2.005	1.9775
H200004016	PARP16			2.4133	2.2154
H200000467	SLC18A2			2.52	1.9316
H200005857	VLDLR			3.4439	3.134
H200004390	DMC1				2.0052
H200012875	KIF1B				2.5715
H200008545	TMEM9				2.192
H200004439	FAM89A				2.3003
H200016561	PAG1				2.8081
H200004662	PELI2				2.0424
H200012241	TMEM154				2.2427
H200009493	CYB5R1				2.5233
H200018222	GPR84				2.3565
H200001460	SLC2A3				2.8812
H200018921	OSBPL6				2.0907
H200006320	SDS				2.0108
H200002343	MASTL				2.0644
H200010023	CPEB4				2.2967
H200004544	SERPINB8				2.8708
H200000525	PTX3				3.1971