Cytotoxicity

Compound cytotoxicity was assessed after exposure of MDA-MB-231 and A549 cells to a concentration range of paclitaxel (0–100 nM for 72 hours). In both cell lines, a similar toxic response was detected by both SRB and xCELLigence, with comparable IC₅₀ values of 4.78±0.90 nM and 6.44±1.90 nM respectively (t-test, p=0.244) in MDA-MB-231 cells. The normalized toxic response correlated highly between SRB-based and xCELLigence-based detection for both cell lines (Spearman's ρ=0.970 and 0.976 for MDA-MB-231 and A549 resp.) (Fig. 2C).

Cell migration

Conventional Transwell plates have been organized in a sequential setup to perform dynamic observations of cancer cell migration in a time-dependent manner, yielding a series of data similar to the xCELLigence system (Fig. 3A, B). Both chemotactic migration to medium containing 10% FBS and random migration with SF medium on either side of the membrane have been considered. High correlation values for both cell lines were obtained when comparing area calculation and OD with xCELLigence Cell Index (CI) (Fig. 4A). Importantly, it must be noted that correlation coefficients were calculated for overall mean values resulting from three independent experiments, each performed in duplicates for all time points. As described in the Materials and Methods section, all raw data obtained were normalized to the single maximal value over three experiments per quantitation method (CI, pixel area or OD) to eliminate units of measurement. Subsequently, values from random (SF) migration were subtracted from chemotactic (FBS) migration per time point to generate signals of net chemoattraction (Fig. 4B). Pixel area calculations, averaged over three fields per insert membrane (Fig. 4C), correlated with xCELLigence CI measurements for both MDA-MB-231 and A549 cells (Spearman's ρ=0.90 for both cell lines). However, OD measurements showed even stronger correlations with xCELLigence data (Spearman's ρ=0.96 and 1.00 for MDA-MB-231 and A549 resp.). Area calculation correlated with OD measurements in a similar fashion as with xCELLigence for both MDA-MB-231 and A549 (Spearman's ρ=0.89 and ρ=0.90 resp.) (Fig. 4A).

Open in a separate window

Figure 3

Conventional Transwell design for detection of time-dependent cell migration.

A. A Transwell setup consists of an upper chamber (insert) that is placed onto a lower chamber (well). The insert contains a microporous membrane (8 µm pores) allowing passage of tumor cells. After a period of serum starvation a serum-free cell suspension is seeded in the insert and exposed to medium containing potential chemoattractants (by default: medium+FBS). During incubation at 37°C and 5% CO₂, cells migrate toward the bottom side of the membrane. B. Experimental design to assess time-dependent migratory behavior of cultured cells. Both migration toward FBS-containing medium and baseline migration (toward SF medium, no chemoattraction) as a negative control were included. Two times two 24-well Transwell plates were used to examine migration to FBS (positive control – top row) and baseline migration (negative control – bottom row). At ten time points during a 24-hour incubation period inserts were fixed and stained in duplicate. Two inserts containing cell-free media (grey fill) have been included throughout the experiment and fixed and stained after 12 hours incubation to assess background absorption in optical density (OD) measurements. In addition, to exclude influence of inter-plate variability on observed migration rates, each plate contained duplicate two-hour control inserts.

Open in a separate window

Figure 4

Time-dependent migratory pattern of MDA-MB-231 and A549.

A. MDA-MB-231 (left) and A549 (right) cell migration profiles, detected by Transwell experiments (black) and xCELLigence (red). Graphs represent scaled signals (0–1) of net chemoattraction after subtraction of the random migration signal (empty squares in panel A, B), with associated Spearman's Rho values. All results originate from three independent duplicate experiments ± SD. B. Normalization procedure of migration patterns. Raw data (left panel) were normalized to a (0–1) scale (middle panel) through division of all data by the maximum value obtained in three independent experiments. Subsequently, random migration (SF) signals (triangle markers) were subtracted from the positive (FBS) control counterparts (circle markers) per experiment to obtain a pure chemotactic signal (right panel). Example shown is the migratory pattern of MDA-MB-231 cells estimated by pixel area calculation in three experiments (exp 1 - red, exp 2 - green, exp 3 - black). Triangle and circle markers represent negative (SF) and positive (FBS) control data respectively. C. ImageJ-based picture processing. Original pictures were color thresholded to obtain a binary image displaying cellular content as saturated black areas on a white background. Thresholded images were masked to exclude non-cellular particles from the final area calculation. Pictures shown are migrated MDA-MB-231 cells after four hours (top row) and 16 hours (bottom row) of incubation. D. Migratory behavior of MDA-MB-231 cells toward medium+FBS (positive control – filled squares) and background migration (empty squares) as detected by conventional Transwell experiments at ten time points spread over 24 hours of incubation. Area calculation (left) of stained cells and optical density (OD – middle) were compared to the xCELLigence migration pattern, reconstructed from the original high-resolution plot by extrapolating data from the corresponding time points (right). All results represent original data from three independent duplicate experiments ± SD. Picture string (obj. 2.5×) shows migratory status of MDA-MB-231 cells, stained as described, at five different stages during 24 hours of incubation. E. Same as (D) for A549 cells.

At later time points of the experiments, xCELLigence generated larger variances between intra-experimental replicates when compared to area calculations or OD values derived from classic Transwells (Table S1). Briefly, variance between replicates within one experiment increased over time, creating a funnel shaped time-dependent pattern (Fig. 4D, E). Variance component analysis indeed revealed an increase of intra-experimental variance on xCELLigence data in comparison with Transwell data. However, early (<10 h) pixel area values showed similar degrees of variance. Additional experiments using an identical setup yielded similar results (results not shown).

A significant difference was observed between background (serum-free, SF) signals generated by the three methodologies of cell migration quantitation. These signals derived from random movement of cells without exposure to any chemoattractant and, serving as a negative control, showed a different pattern when generated by xCELLigence or by both quantitation techniques using Transwells (Fig. 5). Paired comparisons between time-dependent tracking of background migration were performed between all techniques using a likelihood ratio test and revealed a significant difference between xCELLigence and pixel area calculation (p<0.001) for both MDA-MB-231 and A549. OD measurements differed significantly from xCELLigence (p<0.001) for MDA-MB-231 cells, but showed a similar pattern when compared with xCELLigence data generated from A549 cells (p=0.22).

Open in a separate window

Figure 5

Time-dependent random migration profile of MDA-MB-231 and A549.

Comparison of random migration signals (negative control – SF) between three quantitation methods: pixel area calculation – black, OD - red, xCELLigence - green. A likelihood ratio test revealed a significant difference in slope between area calculation and OD (p<0.001) and area calculation and xCELLigence (p<0.001) for both cell lines and OD and xCELLigence (p<0.001) for MDA-MB-231 only. OD and xCELLigence slopes did not differ significantly (p=0.22) for A549 cells.

xCELLigence Real-Time Cell Analysis (RTCA): proliferation and cytotoxicity

Experiments were carried out using the xCELLigence RTCA DP instrument (Roche Diagnostics GmbH, Mannheim, Germany) which was placed in a humidified incubator at 37°C and 5% CO₂.

Cell proliferation and cytotoxicity experiments were performed using modified 16-well plates (E-plate, Roche Diagnostics GmbH, Mannheim, Germany). Microelectrodes were attached at the bottom of the wells for impedance-based detection of attachment, spreading and proliferation of the cells. Initially, 100 µL of cell-free growth medium (10% FBS) was added to the wells. After leaving the devices at room temperature for 30 min, the background impedance for each well was measured. Cells were harvested from exponential phase cultures by a standardized detachment procedure using 0.05% Trypsin-EDTA (Invitrogen NV/SA, Merelbeke, Belgium) and counted automatically with a Scepter 2.0 device (Merck Millipore SA/NV, Overijse, Belgium), Fifty µL of the cell suspension was seeded into the wells (20, 40, 80, 100, 200, 400 and 800 cells/well for proliferation, 1000 cells/well for cytotoxicity experiments). The cell concentrations of 20, 100, 200 and 400 cells/well were considered for correlation with the SRB method described below. After leaving the plates at room temperature for 30 min to allow cell attachment, in accordance with the manufacturer's guidelines, they were locked in the RTCA DP device in the incubator and the impedance value of each well was automatically monitored by the xCELLigence system and expressed as a Cell Index value (CI). Water was added to the space surrounding the wells of the E-plate to avoid interference from evaporation. For proliferation assays, the cells were incubated during ten days in growth medium (10% FBS) and CI was monitored every 15 min during the first six hours, and every hour for the rest of the period. Two replicates of each cell concentration were used in each test. For cytotoxicity experiments, CI of each well was automatically monitored with the xCELLigence system every 15 min during the overnight recovery period. Twenty-four hours after cell seeding, cells were treated during a period of 72 hours with paclitaxel (0, 1, 2, 5, 10, 20, 50 and 100 nM) dissolved in phosphate buffered saline (PBS). PBS alone was added to control wells. Each concentration was tested in duplicate within the same experiment. CI was monitored every 15 min during the experiment. Three days after the start of treatment with paclitaxel, CI measurement was ended.

xCELLigence Real-Time Cell Analysis (RTCA): migration and invasion

Cell migration and invasion experiments were performed using modified 16-well plates (CIM-16, Roche Diagnostics GmbH, Mannheim, Germany) with each well consisting of an upper and a lower chamber separated by a microporous membrane containing randomly distributed 8 µm-pores. This setup corresponds to conventional Transwell plates with microelectrodes attached to the underside of the membrane for impedance-based detection of migrated cells. Prior to each experiment, cells were deprived of FBS during 24 hours. Initially, 160 µL and 30 µL of media was added to the lower and upper chambers respectively and the CIM-16 plate was locked in the RTCA DP device at 37°C and 5% CO₂ during 60 minutes to obtain equilibrium according to the manufacturer's guidelines. After this incubation period, a measurement step was performed as a background signal, generated by cell-free media. To initiate an experiment, cells were detached using TrypLE Express™ (Invitrogen, Merelbeke, Belgium), resuspended in serum-free (SF) medium, counted and seeded in the upper chamber applying 3×10⁴ cells in 100 µL. After cell addition, CIM-16 plates were incubated during 30 minutes at room temperature in the laminar flow hood to allow the cells to settle onto the membrane according to the manufacturer's guidelines. To prevent interference from evaporation during the experiments, SF medium was added to the entire empty space surrounding the wells on the CIM-16 plates. Lower chambers contained media with or without FBS in order to assess chemotactic migration when exposed to FBS and background migration to SF medium as a negative control accordingly. Signals representing net chemoattraction were obtained by subtracting background (SF) values from the positive control (medium containing FBS) signals. Each condition was performed in quadruplicate with a programmed signal detection schedule of each three minutes during the first 11 hours of incubation followed by each five minutes for three hours and finally each 15 minutes to 24 hours of incubation.

A protocol identical to the above migration experiments was followed for invasion experiments added with the application of a layer of Matrigel on the upper side of the membranes and dynamic process follow-up during 50 hours. Aliquoted Matrigel (Basement Membrane Matrix, growth factor reduced, BD Biosciences, Erembodegem, Belgium) was thawed overnight on ice and mixed with ice cold SF medium to obtain two dilutions corresponding with ±190 µg/µL (10%, v/v) and ±63 µg/µL (3.3%, v/v) (Table 1). All Matrigel handling materials as well as the sealed packs containing CIM-16 upper chambers were stored ice cold overnight. Establishment of a Matrigel layer on the CIM-16 upper chamber membranes was achieved by adding 50 µL of the dilution sequentially on top of four membranes followed by removal of 30 µL, leaving a total of 20 µL Matrigel dilution. Subsequently, the coated upper chambers were incubated at 37°C to homogenously gelify during a minimum of four hours, followed by addition of 160 µL media to the lower and 30 µL SF media to the upper chambers. Equilibration and cell addition was carried out as described above.

All data have been recorded by the supplied RTCA software (vs. 1.2.1). As described below in the “Prestatistical data processing and statistical analysis” section, original high-resolution data sets generated by xCELLigence were exported to MS Excel and reconstructed at a lower resolution by selecting only the data points corresponding with the respective time points of signal detection by the endpoint methods (Fig. 7). CI-data from cytotoxicity experiments have been normalized using the RTCA software to the last data point prior to treatment start. All other results (proliferation, migration and invasion) are based on raw data without CI-normalization and were processed as described above for comparison with conventional methodology.

Open in a separate window

Figure 7

Prestatistical data processing.

A. Schematic depiction of processing kinetic data generated by SRB, xCELLigence and Transwells. Raw data with high time resolution (filled and empty circles), resulting from independent xCELLigence experiments (1, 2, 3 and grey arrows) are reduced to a lower time resolution by selecting only the data points corresponding with the time points of endpoint detection (filled circles only). Subsequently, data have been normalized by dividing all values by the highest value recorded over all experiments per method, resulting in a modified Y-axis scale that ranges from 0 to 1. Finally, the normalized data have been averaged with calculation of SD for the three independent experiments per method. B. Reduction of high-resolution data, generated by xCELLigence, to a low resolution comparable with data from conventional assays. The example shows migration (left) and invasion (right) of MDA-MB-231 cells through two densities of Matrigel. The ten time points in the Transwell method (black arrows) were selected from the xCELLigence plots (grey and blue) to reconstruct a low-resolution graph (black), directly comparable to the Transwell data. An identical approach was applied for all other processes studied.

SRB assay: proliferation

Cells were harvested from exponential phase cultures by trypsinization, counted and plated in 48-well plates. To determine a proliferation curve and calculation of the doubling time, seeding densities ranged from 100 to 2000 A549 or MDA-MB-231 cells/well. Each concentration was tested six times within the same experiment. General cell culture conditions and culture medium used for this method were similar to those applied for the xCELLigence counterpart experiments, as well as applied cell densities (100, 500, 1000 and 2000 cells/cm²). Every day, one plate was fixed by the first step of the SRB assay: culture medium was aspirated prior to fixation of the cells by addition of 200 µl cold 10% trichloroacetic acid. After one hour incubation at 4°C, cells were washed five times with deionized water and left to dry. After collection of all plates during ten days, the following steps of the SRB test were performed as described previously [13], [14]. Shortly, the cells were stained with 200 µl 0.1% SRB dissolved in 1% acetic acid for at least 15 minutes and subsequently washed four times with 1% acetic acid to remove unbound stain. The plates were left to dry at room temperature and bound protein stain was solubilized with 200 µl 10 mM unbuffered TRIS base (tris(hydroxymethyl)aminomethane) and transferred to 96 wells plates for optical density reading at 540 nm (Biorad 550 microplate reader, Nazareth, Belgium). Cell doubling time was calculated from the exponential phase of the growth curve.

SRB assay: cytotoxicity

Cells were harvested as described above. In order to assure exponential growth during the experiments, seeding density was 10³ A549 cells per well and 10³ MDA-MB-231 cells/well. After an overnight recovery period, treatment with paclitaxel (0–100 nM) dissolved in PBS was started. Control wells were added with PBS. Each concentration was tested six times within the same experiment. After 72 hours incubation with paclitaxel, survival was determined by the SRB assay as described above. IC₅₀ values, representing the drug concentration causing 50% growth inhibition, were calculated using WinNonlin software (Pharsight, Mountain View, USA).

Transwell migration assay

Comparative migration experiments were conducted using a conventional 24-well Transwell system (6.5 mm Transwell® (#3422), Corning®, NY, USA) with each well separated by a microporous polycarbonate membrane (10 µm thickness, 8 µm pores) into an upper (“insert”) and a lower chamber (“well”). After 24 hours of serum deprivation, cells were detached using TrypLE™ Express (Invitrogen, Merelbeke, Belgium), counted and resuspended in media without FBS to obtain equal cell densities (2.1×10⁵ cells/cm²) as applied in the xCELLigence RTCA DP system with respect to the membrane seeding surface of both techniques (classic Transwell® membrane surface 0.33 cm², RTCA DP 0.143 cm²). A volume of 250 µL containing 7×10⁴ cells was plated to each insert and 600 µL medium was added to the wells. For each experiment, both chemotactic migration to medium containing 10% FBS and random migration with SF medium on both sides of the membrane have been assessed in parallel Transwell plates. At ten predetermined time points after incubation start (Fig. 3B), inserts were fixed and stained in duplicates and migration was quantitated using two commonly used methods. At each time point, cells were fixed and stained in a 20% methanol/0.1% crystal violet solution during three minutes at room temperature, followed by washing in deionized water to remove redundant staining [15]. Non-migrated cells remaining at the upper side of the membranes were carefully removed with cotton swabs and inserts were dried in darkness overnight. As fixing and staining was performed per set of two inserts with the remaining inserts to be further incubated until the following time point, inserts to be fixed and stained at a time point were transferred to a companion 24-well plate and the remaining inserts immediately replaced in the incubator. Using this approach unfavorable influences caused by continuous switching incubating cells between 37°C and ambient temperatures could be avoided. The following day stained membranes were pictured in three random non-overlapping fields at 10× objective and 10× eyepiece on a transmitted-light microscope (Leica DMBR, Leica Microsystems GmbH, Wetzlar, Germany) equipped with an AxioCam HRc camera (Carl Zeiss MicroImaging GmbH, Jena, Germany). A first method of quantitation was performed by processing all obtained images using ImageJ software (http://rsbweb.nih.gov/ij/). Each image (Fig. 4C, left) was color thresholded to obtain a binary (black & white – 8 bit) image with cellular material portrayed as saturated black areas (Fig. 4C, middle). As a next step all non-cellular artifacts, predominantly visible shadows of empty pores and debris, were removed from each image by performing the particle analysis function with a mask excluding all particles smaller than 100 to 250 pixels dependent on the experiment (Fig. 4C, right). Masking thresholds were set by comparing binary images with their original phase contrast counterparts [16]. Degree of migration for each time point per experiment was determined by calculating the average pixel area of the three fields in duplicate. Inserts were subsequently submerged in 300 µL 1% SDS/1× PBS in order to lyse migrated cells and extract crystal violet stain [17]. Submerged inserts were incubated in darkness overnight on a plate shaker at medium speed to ensure complete lysis. The following day 200 µL of each lysate was transferred to a 96-well plate for optical density (OD) measurement at 590 nm using a Powerwave X microplate scanning spectrophotometer (Bio-Tek, Bad Friedrichshall, Germany), representing a second cell migration quantitation method. Cell-free inserts containing only medium had been included in duplicate throughout each experiment as OD background controls. Reported OD data represent average background-corrected values ± SD obtained from three independent experiments in duplicate.

Transwell Matrigel Invasion assay

Reference cell invasion experiments were carried out using a Transwell plate system as described for migration experiments, added with the application of Matrigel as extracellular matrix component. Matrigel dilutions were prepared as described above for the xCELLigence RTCA invasion assay. In order to obtain Matrigel surface area densities synchronized with the CIM-16 plates used for xCELLigence, two dilutions of 20% and 7.7% (v/v) have been prepared in ice cold SF medium corresponding with ±190 µg/µL and ±63 µg/µL respectively, as applied in a volume of 20 µL per insert membrane, identical to the CIM-16 upper chamber coating volume (Table 1). All other conditions regarding culturing, cell seeding density and serum deprivation were identical to the Transwell migration assays described above. At a panel of ten predetermined time points, inserts were fixed and stained in duplicates and invasion was quantified by OD reading at 590 nm after overnight extraction of the crystal violet stain. Reported OD data represent average background-corrected values ± SD obtained from three independent experiments in duplicate.

We thank Ken Op de Beeck (CORE and Center for Medical Genetics, University of Antwerp) for assistance with cell line authentication, the Laboratory for Pathophysiology (University of Antwerp) for granting access to the microscope for picturing Transwell membranes and the Laboratory for Experimental Medicine and Pediatrics (University of Antwerp) for granting access to the microplate scanning spectrophotometer for OD measurements. We gratefully acknowledge Marc Baay and Johan Ides (CORE, Antwerp) for constructive discussions.