Cell type specific applicability of 5-ethynyl-2′-deoxyuridine (EdU) for dynamic proliferation assessment in flow cytometry



Using the nucleoside analogue EdU (5-ethynyl-2′-deoxyuridine) for thymidine substitution instead of BrdU (5-bromo-2′-deoxyuridine) in cell proliferation assays has recently been proposed. However, the effect of EdU on cell viability, DNA synthesis, and cell cycle progression and consequently its usability for dynamic cell proliferation analysis in vitro has not been explored. We compared the effect of EdU and BrdU incorporation into SK-BR-3 and BT474 breast cancer cells and the impact on cell cycle kinetics, cell viability, and DNA damage. We found that EdU can be used not only for pulse but also for continuous cell labeling and henceforth in high resolution EdU/Hoechst quenching assays. BrdU and EdU proliferation assays based on click chemistry revealed comparable results. However, cell viability of SK-BR-3 breast cancer cells was highly affected by long term exposure to EdU. Both SK-BR-3 as well as BT474 cells show cell cycle arrests upon long term EdU treatment whereas only SK-BR-3 cells were driven into necrotic cell death by long term exposure to EdU. In contrast BT474 cells appeared essentially unharmed by EdU treatment in terms of viability. Consequently using EdU enables highly sensitive and quantitative detection of proliferating cells and facilitates even continuous cell cycle assessment. Nevertheless, potential cellular susceptibility needs to be individually evaluated. © 2009 International Society for Advancement of Cytometry

Dynamic cell proliferation assessment using flow cytometry is a potent approach for identifying and quantifying the effect and efficacy of, for example, growth factors and anticancer drugs on tumor cells (1). 5-bromo-2′-deoxyuridine (BrdU)-based techniques constitute powerful tools to determine cell cycle kinetics and to disclose potential mitogenic, cytostatic, and cytotoxic effects upon specific cell treatment (2, 3). Usually, BrdU incorporation into DNA is detected either by quenching of the DNA binding Hoechst 33258 fluorescence (BrdU/Hoechst quenching technique) (4, 5) or by antibody-based BrdU detection (6, 7). Both approaches require a stoichiometric DNA counterstaining.

Antibody-based detection of the thymidine analogue BrdU demands the often tricky and elaborate DNA denaturation facilitating sterical access of antibodies (8). Alternatively, 5-ethynyl-2′-deoxyuridine (EdU), structurally similar to the natural nucleoside, can be coupled via click chemistry (9, 10). EdU detection is based on a copper-catalyzed covalent reaction between a dye-conjugated azide and the alkyne group of the EdU (Fig. 1). The small sized dye-azide allows for efficient EdU detection upon incorporation using gentle conditions. Aldehyde-based fixation and detergent permeabilization for the dye (e.g. Alexa Fluor® dye)-conjugated azide enables access to the alkyne group of EdU. In biological systems, the alkyne is unreactive. Nevertheless, the potential impact EdU on cell viability, DNA synthesis, and cell cycle progression needs to be explored.

Figure 1.

DNA labeling of proliferating cells using the Click-iT® EdU Cell Proliferation Assay (Molecular Probes/ Invitrogen, OR). The EdU nucleoside analogue contains an alkyne that reacts in a copper-catalyzed reaction with an azide (click chemistry) forming a very stable covalent bond. In contrast to antibody-based detection of BrdU labeled DNA prior denaturation of dsDNA is not required.

The flow cytometric proliferation assessment based on BrdU/Hoechst quenching enables the analysis of cell cycle progression with high time resolution (4, 5, 11, 12). Here, we show for the very first time that quenching of the Hoechst 33258 fluorescence is not only a feature of BrdU but also of EdU. However, both the BrdU- and EdU-based Hoechst quenching requires continuous cell incubation with the nucleoside analogue and any potential cytostatic effect has to be identified (4, 11, 12).

It has been shown for halogenated deoxyuridine that substitution of thymidine provokes conformational alterations of the 3D structure of the DNA that can be recognized by DNA binding and/or interacting molecules altering the stickiness of these proteins to the DNA both affecting gene expression and DNA replication (13). An increase in the DNA melting temperature upon nucleoside analogue incorporation has also been described to hamper DNA uncoiling essential for DNA replication and in turn to potentially result in cell death (14). In addition, exposure of cells to nucleoside analogues, namely BrdU, possibly disturbs the natively balanced nucleotide pool. The incubation in the presence of 2′-deoxycytidine (DC) has been shown to reverse a potential disequilibrium (15). Nevertheless, blockage of cell cycle progression (16), induction of differentiation-like events (12), and even induction of cell death by cell exposition to nucleoside analogues have been described (17). Cell susceptibility to nucleoside analogue-mediated toxicity or cell cycle inhibition may be due to hyperextensive import or an inefficient export machinery (18). Additionally, nucleoside incorporation into DNA can be performed with different efficiency due to a distinct coexpression pattern of the DNA-synthesizing machinery in cell lines tolerating the substituted nucleosides to a different extent (13). As a result, nucleoside analogues in DNA incorporation may lead to impaired DNA replication that finally results in single and/or double DNA strand breaks. Usually damaged DNA is in turn recognized and repaired while cell cycle progression is stopped. If the DNA damage is severe, for example, due to a significant amount of strand breaks, cells may undergo apoptosis to obviate further cell cycle progress of cells carrying DNA damages. However, in cancer cells apoptotic pathways or signaling that causes cell cycle arrest are often defective, for example, by mutations in the tumor suppressor gene coding for p53 (17, 19) or additional proteins involved in posttranscriptional mismatch (20) and double strand repair (21). Severe DNA damage represented by double-strand breaks in chromatin triggers the phosphorylation of Histone H2AX on Ser139 which can be used to estimate the amount of DNA damage upon treatment with nucleosides (22).

In consideration of potential undesired effects caused by EdU, we evaluated possible impacts of cell treatment with EdU on several cellular parameters using EdU concentrations covering the range of 0.2 μM–20 μM. We analyzed both pulse and long-term exposure of SK-BR-3 and BT474 breast cancer cell lines to the nucleoside analogue. Cell proliferation was dynamically monitored and dead vs. viable cell fractions were quantified upon EdU treatment. The annexin-V-FITC/propidium iodide (PI) assay was utilized to distinguish between apoptotic and necrotic cell death. H2AX phosphorylation was imaged to identify EdU-mediated DNA damages.

We found BT474 and SK-BR-3 cells differentially sensitive to EdU treatment. However, using low concentration EdU for pulse labeling click chemistry-based cell preparation was much easier and faster to perform, and subsequent evaluation of cell cycle kinetics turned out to be superior to BrdU labeling in terms of sensitivity. Using EdU (instead of BrdU) facilitates a highly reliable cell proliferation assessment in vitro and moreover, uncloses advanced proliferation analysis including multiplexed measurements (23).


Cell Culture

The human breast cancer cell lines SK-BR-3 and BT474 were obtained from the American Type Culture Collection (ATCC) and grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal calf serum (FCS) (both from PAN Biotech, Aidenbach, Germany). 5% instead of 10% FCS was used to minimize potential stimulating (or inhibiting) effects provoked by poorly defined serum ingredients. Cell culture procedures are described in detail in (2).

Cell Treatment with EdU and BrdU and Annexin V/Propidium Iodide Assay

For cell incubation with EdU and BrdU, 1.66 × 105 cells were seeded (day one) into T25 culture flasks in DMEM medium supplemented with 5% FCS. Cells were treated on day 4 (96 h incubation interval) or on day 6 (48 h incubation interval) with 0.2 μM, 1 μM, 10 μM, or 20 μM EdU (5-ethynyl-2′-deoxyuridine, Molecular Probes/Invitrogen, OR) either alone or in combination with half-equimolar 2′-deoxycytidine (DC, Sigma-Aldrich, Taufkirchen, Germany) to avoid a potential thymidine analogue-induced inhibitory effect (4, 15). Similar concentrations were used for 5-bromo-2′-deoxyuridine (BrdU, Sigma-Aldrich) and DC treatment. For short term incubation, nucleosides were thoroughly removed after 3 h (EdU and EdU/DC incubation) or 30 min (BrdU/DC incubation) by three wash steps with PBS. For long term incubation, nucleoside analogues were not removed until harvesting. Medium (and nucleosides in the case of long term nucleoside incubation) was replaced on day 6. Cells were harvested on day 8 and used for live/dead labeling and EdU labeling with click chemistry as described below.

For analysis of apoptosis, the cell culture supernatant of each sample was saved and merged with harvested cells. Then, samples were labeled with annexin V/propidium iodide using the TACS™ Annexin V-FITC kit (R&D Systems, Wiesbaden-Nordenstadt, Germany) as described in (2).

As it was shown that incorporated BrdU sensitizes cells to short-wavelength light, all cell culture flasks were handled without direct light irradiation and cell culture flasks were wrapped in aluminum foil for incubation. According to the manufacturers recommendation, we also protected EdU-labeled cells from direct light irradiation in the same way.

Anti-BrdU Technique, BrdU/Hoechst, and EdU/Hoechst Quenching Technique

Cell preparation and antibody staining for detection of BrdU incorporation with the anti-BrdU technique and the BrdU/Hoechst quenching technique was performed as described in detail previously (2, 24) and adopted for evaluation of EdU/Hoechst quenching.

Live/Dead Cell Labeling

Live/dead cell labeling was done with the LIVE/DEAD® Fixable Dead Cell Stain Kit (Molecular Probes/Invitrogen, OR) according to the manufacturer's instructions. Briefly, 1 × 106 cells were washed twice in PBS, resuspended in 1-ml PBS containing 1 μl LIVE/DEAD® Fixable Violet dead cell stain and incubated for 30 min on ice. After washing once in PBS, cells were fixed in 1 ml formaldehyde (4%, SG Planung, Holzkirchen, Germany). After 15 min on ice, cells were washed once in PBS/1% BSA and transferred to ice until analysis with a BD™ LSR II four laser flow cytometer (BD Biosciences, San Jose, CA).

Determination of EdU Incorporation with Click Chemistry

Cells were seeded, pulse-labeled with EdU for 3 h, and harvested as described earlier. EdU concentrations for pulse labeling were 10 μM and 0.1 μM. Nontreated cells served as control. Detection of EdU incorporation into the DNA was performed with the Click-iT® EdU Alexa Fluor® 488 Cell Proliferation Assay Kit (Molecular Probes, Invitrogen, OR, USA) according to the manufacturer's instructions. In brief, 1 × 106 harvested cells were washed twice in PBS/1% BSA and fixed in 100 μl Click-iT® fixative. After a 15 min incubation step at room temperature in the dark, cells were washed twice in 1× saponin-based permeabilization and wash reagent. The Click-iT® EdU reaction cocktail (1×) was prepared according to the manufacturer's instructions and added to the cell pellet (500 μl per 1 × 106 cells). Samples were incubated for 30 min at room temperature in the dark and washed with 1× saponin-based permeabilization and wash reagent.

For staining of cellular DNA, samples were washed once in 1× saponin-based permeabilization and wash buffer. The DNA staining solution was prepared with 50 μl saponin-based permeabilization and wash buffer (10×), 5 μl RNase, 2 μl Click-iT® EdU CellCycle 633-red stain and 450 μl PBS/1% bovine serum albumin (BSA, 22%, Biotest AG, Dreieich, Germany) per sample and added to the cells. After an incubation period of 15 min at 37°C in the dark, cells were transferred to ice until analysis with a BD FACSCalibur™ flow cytometer (BD Biosciences, San Jose, CA).

Cell Counting and Univariate Cell Cycle Assessment after 96 h and 144 h EdU Treatment

SK-BR-3 and BT474 cells were seeded in Petri dishes at a cell density of 1.4 × 105 on day 1 and treated with 0.2 μM EdU on day 3 for 144 h or day 5 for 96 h. Medium supplemented with EdU was replaced every 2 days. Cells were harvested on day 9 and the cell number was obtained by counting in a Neubauer cell counting chamber (Paul Marienfeld GmbH, Bad Mergentheim, Germany). Cells were stained with PI (final concentration 25 μg/ml) for conventional univariate DNA labeling as described in (2) and measured on a BD FACSCalibur™ flow cytometer.

Flow Cytometry Instrumentation and Data Analysis

Univariate DNA staining, evaluation of apoptosis as well as EdU measurements were done on a standard BD FACSCalibur™ flow cytometer. FITC and Alexa Fluor® 488 signals were measured on FL1 with a 530/30 bandbass filter, PI on FL3 with a 670 nm longpass filter. A BD™ LSR II four laser flow cytometer was used for live/dead discrimination. The violet laser line (405 nm) was used for violet-fluorescent reactive dye excitation. Fluorescence emission was detected via 450/50 bandpass filter optic. Sample measurements were performed with CellQuest™ Pro Software (BD Biosciences, San Jose, CA) using a Macintosh G4 computer for the BD FACSCalibur™ flow cytometer and with FACSDiva™ Software (Version 5.0.1, BD Biosciences, San Jose, CA) on a personal computer for measurements on the BD™ LSRII flow cytometer. Data analysis was performed with FlowJo Software (Ashland, OR). Cell debris and aggregates were excluded from analysis using pulse processing (DNA width signal against DNA area signal). For static DNA analysis, live cells were gated with Expo 32™ v1.2 software (AppliedCytometry, Sheffield, UK) and used for quantification of cell cycle distribution with WinCycle software (Phoenix Flow Systems, San Diego, CA).

High Throughput Imaging Analysis

Cells were plated in 96 well flat bottom plates (Greiner Bio-One, NC) (3,000 cells per well). After cell growth for 24 h, DMEM containing 20 μM BrdU/10 μM DC, 2 μM BrdU/1 μM DC, 20 μM EdU, 2 μM EdU, 0.2 μM EdU, or DMSO as control was added to half of the plate. Control cells and cells treated with 5 μM etoposide (Sigma-Aldrich, St. Louis, MO) served as negative and positive control, respectively, for the induction of double strand breaks. After an incubation period of 24 h, medium supplemented as indicated above was replaced in the entire plate and incubated for additional 24 h. This treatment protocol results in one half of the plate incubated for 48 h and the other half treated for 24 h. Subsequently, the supernatant was removed and cells were washed once with D-PBS (Dulbecco's Phosphate-Buffered Saline, Molecular Probes/Invitrogen, OR). After fixation in 3.7% formaldehyde (Sigma-Aldrich, St. Louis, MO) in D-PBS containing Ca2+ and Mg2+ (Molecular Probes/Invitrogen, OR) for 15 min, cells were washed with 1% BSA/D-PBS and fixed with 0.5% Triton® X-100 (Sigma-Aldrich, St. Louis, MO) in D-PBS/Ca2+/Mg2+ for 20 min. After a washing step with 1% BSA/D-PBS, nonspecific binding sites were blocked by incubation with 3% BSA/D-PBS overnight at 4°C. Next day, the blocking buffer was aspirated. For phospho-H2AX labeling, the primary antibody (clone JBW301, Millipore, Billerica, MA) was diluted to 5 μg/ml in 3% BSA/D-PBS and added to the cells. After an incubation period of 1 h, cells were washed with 3% BSA/D-PBS and incubated with Alexa Fluor® 488 goat anti-mouse IgG (H+L) secondary antibody (Molecular Probes/Invitrogen, OR) at 2 μg/ml in 3% BSA/D-PBS for 30 min. Cells were washed once in 3% BSA/D-PBS and twice in D-PBS. DNA staining was performed with 20 μg/ml Hoechst 33342 in D-PBS for 15 min. After three wash steps in D-PBS cells were imaged on an ArrayScan® VTI HCS reader (Cellomics Inc, Pittsburgh, PA). Data analysis was performed with Panmo (Chi-Square Works Inc, Seabeck, WA) and graphing was done using Prism (GraphPad Software, San Diego, CA).

Immunochemical Anti-Cyclin B1 Staining and Microscopic Analysis

SK-BR-3 and BT474 cells were treated with 0.2 μM EdU for 144 h, harvested and fixed in 4% formaldeyde. Cells were prepared with the Shandon Cytoblock Cell Block Preparation System (Thermo Electron Corporation, Cheshire, United Kingdom) and embedded in paraffin (Sigma-Aldrich Chemie GmbH, Deisenhofen, Germany) in a HyperCenter Excelsior (Thermo Fisher Scientific GmbH, Dreieich, Germany). For immunochemical staining, slides were dried and deparaffinized in a descending alcohol series. Demasking of binding epitopes for the primary antibody was performed by microwave treatment at 100°C in citrate buffer for 30 min. Slides were cooled down by incubation in an ice bath for 20 min. After five times washing in distilled water, immunochemical staining was performed with the automated IHC slide staining system NexES® IHC with mouse monoclonal anti-CycB1 antibody (Clone 7A9, Novocastra, Newcastle upon Tyne, UK) in a dilution of 1:10. Slides were imaged with an Axio Imager Z.1 using AxioVision 4.5 software.

Statistical Analysis

The Student-t-test was applied to compare means of two independent samples. Each measurement was performed at least in triplicate (n = 3).


Applicability of EdU as a BrdU Replacement

EdU was used as a replacement for BrdU for dynamic cell proliferation assessment of BT474 breast cancer cell line both for the pulse labeling anti-BrdU technique (Fig. 2A or 2B) and for the continuous labeling BrdU/Hoechst quenching technique (Fig. 2C or 2D). The separation of nucleoside analogue labeled S-phase and nucleoside analogue negative G1- and G2/M-phase cells is similar for BrdU detection with anti-BrdU antibodies (49-fold difference) and labeling EdU with click chemistry (55-fold difference) at the 0 h time point (Fig. 2A or 2B). The BrdU/Hoechst quenching technique allows to dynamically assess cell proliferation and to discriminate cells in different cell cycle phases of subsequent cell cycles (4). EdU is capable of quenching Hoechst 33258 to a similar extent when compared with BrdU (Fig. 2C or 2D). One representative time point (39.5 h) is shown. The time course of EdU quenching with cells continuously incubated with EdU for different time intervals is similar to the time course with BrdU (data not shown).

Figure 2.

Comparison of the applicability of BrdU/DC and EdU for dynamic cell proliferation assessment in BT474 cells. A,B: anti-BrdU vs. click chemistry detection of EdU. Cells were pulse labeled with 20 μM BrdU/10 μM DC for 30 min (A) or with 20 μM EdU for 3 h (B). DNA labeling was performed with PI for BrdU-(A) and CellCycle 633 Dye for EdU-labeled cells (B). C, D: BrdU/Hoechst quenching vs. EdU/Hoechst quenching. Cells were continuously exposed to 120 μM BrdU/60 μM DC (C) or 100 μM (D) EdU for 39.5 h and labeled with Hoechst 33258 and PI as described in Materials and Methods.

Cell Death Induced by Long Term EdU Treatment

Live/dead discrimination

The live/dead discrimination assay is based on binding of the reactive dye to free amines on the cell surface of a live cell and to free amines in the cell's interior, if the membrane is compromised. Therefore, staining of live cells results in a dim stain in contrast to dead cells that show highly intense staining (Fig. 3A).

Figure 3.

Analysis of cytotoxicity by quantifying the dead cell fraction after live/dead cell gating (A) and annexin V-FITC staining (B and C). A: SK-BR-3 and BT474 cells were continuously labeled with the indicated nucleosides for 48 h or 96 h. Representative histograms with live/dead discrimination of SK-BR-3 cells continuously labeled for 96 h with 20 μM EdU, 20 μM EdU/10 μM DC, or 20 μM BrdU/ 10 μM DC and stained with LIVE/DEAD® Fixable Violet dead cell stain are shown. M1: live cells, M2: dead cells. Differences in dead cell fractions are obvious however do not represent a significant deviation. B: Necrotic cell death induced by EdU and EdU/DC determined with the annexin V-FITC/PI Assay. SK-BR-3 cells were continuously labeled with nucleosides for 96 h. C: Kinetics of cell death in SK-BR-3 mediated by continuous incubation with 20 μM EdU. Positive controls: 10 μM Camptothecin, 14.5 h; negative controls: nontreated cells, 96 h BrdU/DC-treated cells.

The fraction of dead SK-BR-3 breast cancer cells increases in a concentration dependent manner upon long term treatment both with EdU and with EdU/DC (Fig. 3A). The dead cell fraction for EdU treatment increases from 8.6% ± 4.5 (control cells) to 20.5% ± 1.2 (0.2 μM), 27.9% ± 7.0 (1 μM), 43.2% ± 13.7 (10 μM), and 44.3% ± 17.3 (20 μM) after 96 h. The amount of cell death upon EdU treatment is not rescued by the addition of DC. Only minor effects on the amount of cell death are seen with BrdU/DC after both 48 h and 96 h and with EdU after 48 h. BT474 seems to be unaffected by treatment with EdU, EdU/DC, and BrdU/DC both after 48 h and 96 h of incubation.

Pulse labeling for short durations with EdU, EdU/DC, or BrdU/DC has no effects on the dead cell fraction, both in SK-BR-3 and in BT474 (data not shown).

Annexin V/propidium iodide assay

Annexin V-FITC/PI labeling was performed to analyze the mode of EdU-mediated cell death in more detail and to discriminate between apoptotic and necrotic cell death. EdU induces SK-BR-3 cell death by necrosis rather than by apoptosis in a concentration (Fig. 3B; copositive for annexin V-FITC and PI: upper right quadrant) (25) and time (Fig. 3C)-dependent manner. Although after 96 h of continuous nucleoside exposure control cells show a 2.1% (control cells) and 2.3% (BrdU/DC-treated) (4) necrotic cell fraction (Fig. 3B), 1 μM EdU induces 22.4% necrotic cell death in the SK-BR-3 cell population, which further increases to 46.3% with 10 μM EdU and to 84.7% with 20 μM EdU. Concomitant addition of half-equimolar DC in combination with EdU does not rescue from EdU-mediated necrotic cell death. The addition of EdU and DC in combination more efficiently increases the necrotic cell fraction when compared with EdU treatment alone. With EdU/DC treatment, the necrotic fraction rises to 29.8% (0.2 μM EdU/0.1 μM DC), 42.1% (1 μM EdU/0.5 μM DC), and 75.4% (10 μM EdU/5 μM DC) finally reaching a slightly lower level of 80.6% with 20 μM EdU/10 μM DC when compared with EdU only treatment (84.7%).

The time-dependent course of cell death was analyzed to prove that EdU mediates cell death via necrosis rather than apoptosis. Figure 3C shows the EdU mediated, time dependent increase of the necrotic cell fraction of SK-BR-3 that were exposed to 20 μM EdU for different time intervals. The necrotic cell fraction significantly rises from 2.2% of control cells (without EdU) to 7.6% after 48 h and advances to a level of 47.5% after 96 h. The increment in the necrotic cell fraction is accompanied by a less pronounced increase in the apoptotic cell fraction from 3.6% (without EdU) to 19.1% (96 h). The live cell fraction falls from 91.5% (without EdU) to 30.9% after 96 h incubation with EdU.

For a positive control, SK-BR-3 cells were treated with 10 μM Camptothecin for 14.5 h (Fig. 3B or 3C) (2). 34.8% of SK-BR-3 cells undergo apoptotic cell death by treatment with Camptothecin which is represented by annexin V-FITC staining before showing membrane permeability by PI staining (Fig. 3B). At this time point, 28.3% of SK-BR-3 cells are found in the upper right quadrant representing late apoptotic/necrotic cells. BrdU/DC does not induce cell death in SK-BR-3 (Fig. 3B or 3C).

Determination of the Origin of Dead Cells

Figure 4 shows the origin of necrotic cells upon 96 h EdU treatment by EdU/Hoechst quenching. The main track of dying cells derives from G1-phase of the third cell cycle, that is, not before two cell divisions were completed (1, 4, 5, 11, 12, 16).

Figure 4.

EdU/Hoechst quenching measurement of SK-BR-3 cells after 96 h EdU treatment. The Dot Plot exemplifies the exit of necrotic cells upon bifiliary EdU incorporation (1, 4, 5, 11, 12, 16). Damaged cells (see rectangle) predominantly derive from G1-compartment of the third cell cycle (arrow) after two cell cycles are completed.

Uptake Efficiency of EdU into DNA

To investigate the causes for the different sensitivity of SK-BR-3 and BT474 to EdU-mediated necrotic cell death, we analyzed EdU incorporation efficiency of both the cell lines. To determine whether or not the lower toxic effect of EdU in BT474 cells was due to a more efficient nucleoside export from BT474 cells which in turn would result in a less efficient EdU incorporation into the DNA (when compared with SK-BR-3 cells), both cell lines were pulse labeled with 1/100 of the recommended EdU labeling concentration (0.1 μM EdU) and EdU labeling density was measured with click chemistry. As a control, the 10 μM EdU concentration (recommended for efficient DNA labeling) (Click-iT® EdU Flow Cytometry, Invitrogen MP, Assay Kits, 2007) was used.

With 10 μM EdU treatment of SK-BR-3 and BT474 and EdU labeling with the Click-iT® EdU Alexa Fluor® 488 Cell Proliferation Assay Kit, EdU-positive S-phase cells (MFI: 1370 in SK-BR-3 and 1312 in BT474) widely separate from EdU-negative cells (MFI: 18.2 in SK-BR-3 and BT474, Fig. 5A or 5B). Even upon EdU labeling using just 0.1 μM EdU, both EdU-positive SK-BR-3 and BT474 cells could easily be identified by a significant shift to higher MFI values (272 and 558, respectively) (Fig. 5). Staining intensity of EdU-positive cells is lower when compared with a labeling based on a concentration of 10 μM EdU. Nevertheless, the separation of EdU-positive and -negative cells shows to be similar in SK-BR-3 and BT474 (78.9% vs. 83.7% decrease in the MFI ratio of EdU positive/EdU negative cells from 10 μM to 0.1 μM EdU treatment, respectively) indicating that the incorporation efficiency of EdU at both the concentration of 10 μM and 0.1 μM is similar in both cell lines.

Figure 5.

Evaluation of EdU uptake into DNA. SK-BR-3 and BT474 cells were pulsed with 0.1 μM low and 10 μM standard EdU concentrations for 3 h and subsequently harvested. EdU was detected with the Click-iT® EdU Alexa Fluor® 488 Cell Proliferation Assay Kit (A). Histograms of Click-iT® EdU Alexa Fluor® 488 stained cells were overlayed. The ratios of the mean fluorescence intensities of EdU-positive and EdU-negative cells decrease from 10 μM EdU-treated cells to 0.1 μM treated cells to a similar extent in BT474 and SK-BR-3.

Cell Cycle Distribution upon EdU Treatment

Long term (48 h, 96 h) EdU treatment at different concentrations (0.2 μM–20 μM, data not shown) had diverse effects when compared with control cells depicting the dynamics of cell proliferation. This complex performance of EdU is exemplified in Figure 6A by an increase in S-phase fraction (SPF) for BT474 cells after 96 h 0.2 μM EdU treatment (25.5%) when compared with control cells (19.5%), whereas after further EdU incubation (144 h), this trend is reversed and SPF is decreased (16.9%). In contrast in SK-BR-3, SPF increased in a time-dependent manner from 20.6% in control cells to 25.1% in 96 h 0.2 μM EdU-treated cells and to 32.5% at the 144 h time point. Although an increase in SPF is generally regarded as evidence for increased cell proliferation, cell counting proves that the cell cycle disturbance is accompanied by a reduction in cell multiplication after EdU treatment. The cell number is decreased in BT474 from 5.5 ± 1.5 × 105 cells in the nontreated sample to 1.8 ± 0.3 × 105 cells after 144 h 0.2 μM EdU treatment and in SK-BR-3 from 8.9 ± 1.3 × 105 cells (control) to 2.3 ± 0.03 × 105 cells (144 h 0.2 μM EdU).

Figure 6.

A: EdU mediated G2/M arrest. Cell cycle distribution upon treatment with 0.2 μM EdU for 96 h and 144 h and respective DNA histograms are shown. Significant differences in cell cycle phases compared to the control are indicated by an asterisk. Corresponding cell numbers are indicated. B: Immunohistochemical Cyclin B1 staining of nontreated (−) and 0.2 μM EdU treated (+) SK-BR-3 and BT474. Cytoplasmic Cyclin B1 accumulation under EdU treatment, which is coincident with a G2/M arrest (A) is observed in both cell lines (BT474: 30% w/o EdU vs. 60% with EdU) (SK-BR-3: 30% w/o EdU vs. 50% with EdU).

Inhibition of cell cycle progression by long time EdU treatment has been proven by a decrease in G1-phase fraction in both cell lines over time with 0.2 μM EdU treatment (BT474: control: 74.7%, 96 h EdU: 53.3%, 144 h EdU: 53.3%; SK-BR-3: control: 68.3%, 96 h EdU: 53.1%, 144 h EdU: 24.5%) and a time-dependent increase in G2/M-fraction (BT474: control: 5.9%, 96 h EdU: 21.2%, 144 h EdU: 50.3%; SK-BR-3: control: 11.1%, 96 h EdU: 21.9%, 144 h EdU: 43.2%).

In contrast, BT474 and SK-BR-3 cells treated with a short EdU pulse (3 h) showed no change in SPF (data not shown). In addition, short time exposure of five different cell lines (HeLa, Jurkat, U266, 3T3, CHO) to EdU and BrdU showed no effect on cell viability or cell cycle distribution in three phases of cell cycle when compared with the control (data not shown).

Cyclin B1 Expression upon EdU Treatment

In addition to flow cytometric quantification, we immunochemically stained BT474 and SK-BR-3 cells against Cyclin B1 expression (Fig. 6B) which, if cytoplasmically located, is indicative for G2-phase. We found in both cell lines a significant increase of Cyclin B1 positive cells under EdU treatment substantiating flow cytometrically observed G2-phase arrest. Counting three microscopic visual fields of Cyclin B1 positive and negative cells, we found 60% positive cells under EdU treatment vs. 30% without EdU treatment in BT474 and 50% positive SK-BR-3 vs. 30% negative cells, respectively.

Determination of Phosphorylation of Histone H2AX

To determine the degree of double strand DNA breakage as measured by a positive phospho-H2AX signal, high throughput imaging analysis was used. Long term continuous incubation of EdU on SK-BR-3 breast cancer cell line shows a dose and time-dependent increase of phospho-H2AX when compared with the control (Fig. 7). For 48 h incubation, the signal increased from 13.7% ± 1.5% (0.2 μM EdU) to 29.1% ± 1.8% (2 μM EdU) and 38.3% ± 3.0% (20 μM EdU) when compared with 5.6% ± 0.7% (control cells). The highest EdU concentration resulted in a similar signal upon the short term (24 h) exposure to 5 μM etoposide (39.0 ± 1.9%), which is known to inhibit (or even to block) cell cycle progress by causing double-stranded DNA breaks. At the shorter exposure of 24 h to EdU, DNA damage is evident and significantly above the nontreated control. In contrast BT474 cells have approximately one half of the phospho-H2AX signal at the higher EdU concentrations but are still above the nontreated control (5.6 ± 0.7% vs. 10.2 ± 1.6% (0.2 μM), 16.5 ± 1.1% (2 μM), and 17.2 ± 1.3% (20 μM EdU), Fig. 7). Continuous BrdU incubation in combination with half-equimolar DC in SK-BR-3 or BT474 (Fig. 7) has no significant increase on phospho-H2AX with 3.6 ± 0.8% in nontreated cells and 2.9 ± 0.7% (2 μM), and 3.9 ± 1.2% (20 μM) in SK-BR-3 cells, and 5.6 ± 0.7% in control cells and 5.8 ± 2.1% (2 μM), and 7.4 ± 1.9% (20 μM) in BT474 cells.

Figure 7.

Analysis of cytotoxicity measured by H2AX phosphorylation as marker for DNA double strand breaks. A: After 48 h EdU causes a concentration dependent H2AX phosphorylation with a higher impact on SK-BR-3 compared to BT474 at similar EdU concentrations. B: Representative examples of microscopic images of 24 h with 20 μM BrdU (left) or 20 μM EdU (right)-treated SK-BR-3. Blue: Hoechst 33342 staining of the nuclei; green: phospho-H2AX staining.


Flow cytometry-based dynamic proliferation assessment is an undisputed powerful approach for high resolution monitoring of cell cycle progression in vitro providing multiplexed information about, for example, cell cycle duration, transition points, proliferative quiescence, etc. (1, 11). The most common approaches are based on DNA labeling of proliferating cells by either pulse exposing the total population either (e.g., 30 min) or continuously (for e.g., 72 h) with a thymidine nucleoside analogue. For this purpose, usually BrdU is used that is added in excess to the culture medium and replaces thymidine during DNA replication. After cell harvesting of numerous samples, the progress of BrdU-labeled cells can be identified and quantified over a predefined period of time by either an anti-BrdU antibody staining (anti-BrdU technique) (6, 7) or by a double DNA staining technique (BrdU/Hoechst quenching technique) (5, 11, 12, 16). These methods clearly extend the potential and the amount of accessible information of conventional proliferation assays (e.g., MTT like assays, H3-thymidine incorporation); however, there is a strong demand for an accurate cell preparation and measurement execution.

In particular, the synchronous antibody-based BrdU detection and total DNA staining demands for achieving a compromise between a sufficient degree of denaturation on one hand and sufficient preservation of double DNA strands on the other hand, a tricky and time consuming approach.

For comparative purposes, we have pulse and continuously labeled BT474 and SK-BR-3 breast cancer cell lines with nucleoside analogues EdU and BrdU and executed preparation, staining, and detection as recommended (6, 11) (Click-iT® EdU Flow Cytometry, Invitrogen MP, Assay Kits, 2007). We found click chemistry-based EdU detection upon pulse treatment to yield nearly the same staining intensity as conventional antibody staining of BrdU positive cells (Fig. 2A or 2B) indicating that EdU and BrdU can be detected with almostthe same sensitivity. The examples shown in Figures 2C or 2D and 4 demonstrate for the first time the usability of EdU for the quenching-based proliferation analysis using Hoechst 33258 and PI for cell cycle recording upon continuous labeling.

Further on, we have found a dose and time-dependent effect of EdU labeling on SK-BR-3 but not on BT474 cell viability (Fig. 3A). In SK-BR-3 cells, EdU incorporation causes a concentration dependent increase of dead cells but not until an incubation period of 96 h. The incubation in DC-supplemented medium does not rescue SK-BR-3 cells from EdU-mediated cell death. Because of the immediate transition of viable cells (annexin V-FITC/PI double neg. cells) to an annexin V-FITC/PI double positive phenotype, the dead SK-BR-3 cells observed after 96 h EdU incubation have to be attributed to a necrotic rather than to an apoptotic fraction (Fig. 3B). The time- and concentration-dependent fraction of apoptotic (annexin V-FITC neg./PI pos.) cells remains unaltered (Fig. 3C). In contrast to the usage of EdU, BrdU does not affect the viability of SK-BR-3 cells.

To exclude that SK-BR-3 cell sensitivity to EdU treatment has to be attributed either to a more efficient EdU uptake, we quantified the amount of nucleoside incorporation represented by the staining intensities of EdU-positive vs. -negative cells upon 0.1 and 10 μM EdU treatment. Figure 5 shows the nearly identical staining intensities of EdU-positive and negative cells in both cell lines indicating that the nucleoside incorporation into the DNA occurs in equivalent quantities. Hence different susceptibilities can not be attributed to a different EdU import or export efficiency. However, this observation is not conclusive with respect to a potential different EdU metabolism in BT474 and SK-BR-3 cells that might be due to a disturbance of a cell's nucleoside pool (15). Applied in excess BrdU was found to be an allosteric inhibitor of ribonucleotide reductase (responsible for cytidine and uridine diphosphate reduction to the deoxynucleotides) (26). Therefore, cells fed with a nucleoside analogue may be driven into cell death by starving them for deoxycytidine nucleotides (3). Supplemented DC was shown to be capable to reverse a toxic effect of BrdU (15), to balance the nucleotide pool and to compensate a potential “thymidine block” (27–29). However, in this study DC was ineffective when EdU was used, indicating that a disturbance of the nucleotide pool is not a major reason for SK-BR-3 cell death mediated by EdU.

Phospho-H2AX detection after EdU treatment was found to be more profound in SK-BR-3 when compared with BT474 cells suggestive for a higher amount of DNA double strand breaks caused by EdU treatment (Fig. 7). DNA fragmentation is observed after 24 h EdU incubation and coincides with the emergence of annexin V-FITC/PI double positive cells, again supporting the interpretation of dead cells as necrotic rather than apoptotic cells (Fig. 3). In consideration of the incubation time of 96 h, it is evident that the induction of necrotic SK-BR-3 cells upon EdU incorporation predominantly occurs not before a bifiliary substitution of thymidine did take place and a certain threshold of thymidine substitution is achieved. This interpretation could be proved by attributing the origin of necrotic cells to G1-phase of the third cell cycle (Fig. 4). Thymidine/thymidine-analogue replacement occurs due to the following substitution ratios: 50, 75, 87.5, and 93.75% substitution after the first, second, third, and fourth cell division, respectively, (1) making it obvious that a unifiliary thymidine substitution after short period of cell exposure to EdU is not sufficient to cause a cell cycle arrest or any significant fraction of dead (necrotic) cells.

The cell accumulation in G2/M-phase after 144 h continuous EdU treatment in 50% of BT474 cells and 43% in SK-BR-3 cells, respectively, that is accompanied by a significant (P = 0.016) increase of S-phase fraction in SK-BR-3 cells (Fig. 5) and cytoplasmic Cyclin B1 accumulation in both cell lines (Fig. 6B) is indicative of an impaired cellular capacity to transition from G2- to M-phase (30). This might be due to functional deficiency of p53 which is known to be different in BT474 and SK-BR-3 cells (19, 31–34). Unaltered p53 is activated upon DNA damage and initiates a set of antiproliferative responses including cell cycle arrest and thereby allows DNA repair before cell replication proceeds. More severe DNA damage causes p53-induced cell death. However, the absence of or mutated p53 can cause uncontrolled cell proliferation without repair as well as reduced apoptosis. SK-BR-3 cells carry a dominant negative mutation in codon 175, exon 5 (CGC to CAC, Arg > His transition), whereas BT474 cells are characterized by another dominant negative mutation in codon 285, exon 8 (GAG to AAG, Glu > Lys transition) (19, 31), which abolish p53 transactivating activity and DNA-binding capacity, respectively (32, 33). The G2/M arrest (BT474 and SK-BR-3) and DNA repair (BT474) are suggestive to account for different p53 defects, that is, the differentially altered p53 protein function (as described elsewhere) (34). Blandino et al. (34) have characterized a gain of mutated p53 function that exerts a direct protective effect and increased drug resistance. Resistance to etoposides has been attributed to codon 175 mutation (BT474) (34), however we found both BT474 and SK-BR-3 cells to be highly susceptible to etoposide treatment as demonstrated by DNA double strand breaks and H2AX phosphorylation (Fig. 7A). In addition, SK-BR-3 cells even exert the apoptotic program as shown previously (2). Resistance to etoposide probably is masked by the etoposide concentration of 5 μM applied here. However, the different sensitivity to EdU might be due to the different gain of function mutations in p53 in both cell lines.

In summary, EdU labeling can be performed rapidly and detection with click chemistry is reliable, highly sensitive, and reproducible. The procedure does not require DNA denaturation, for example, with concentrated hydrochloric acid or methanol/acetic acid mixtures which are most frequently used to identify BrdU-labeled cells. Gentle preparation conditions facilitate the preservation of both protein epitopes and the cells' integrity. Moreover, it enables the experimenter to combine the EdU-based technique with staining of both membrane and intracellular antigens, live/dead cell discrimination, and DNA staining in a multiparametric manner (Click-iT® EdU Flow Cytometry, Invitrogen MP, Assay Kits, 2007) that facilitates to integrate proliferation studies into “cytomics” approaches (23, 35). Because of the excellent tissue penetration of the small-sized fluorescent azides used for EdU detection, this thymidine analogue is particularly suitable for labeling of proliferating cells in organ explants and tissues in vivo under concurrent structural tissue preservation (9).

In a previous study, we have demonstrated that BrdU incorporation into DNA can affect cell cycle progression in vitro (4). However, sensitivity to thymidine analogues turned out to be cell type specific and concentration dependent, however, EdU treatment might properly work instead. Irrespectively, it is obvious that different cell types with different origin differentially and individually respond to nucleoside treatment, an observation that is valid for both types of nucleosides. Consequently, it has to be recommended that in general any potential inhibitory or cytostatic effect caused by both BrdU and EdU treatment (in particular in a continuous treatment setting) needs to be individually evaluated before dynamically assess cell proliferation in vitro.

In conclusion, we have shown that EdU affects the breast cancer cell lines BT474 and SK-BR-3 on two different levels: on the one hand EdU induces cell cycle arrest in G2/M-phase and on the other hand it specifically causes necrotic cell death. However, a short EdU pulse based on 1/200 of the standard BrdU concentration an antiproliferative and/or necrosis inducing impact on SK-BR-3 and BT474 cells can be efficiently avoided. Click chemistry-based detection of EdU incorporation ensures an accurate assessment of cycling cells, basically an approach unclosing advanced proliferation analysis including mutiplexed measurements (23).


The authors thank Elisabeth Schmidt-Bruecken, Rosi Kromas, and Frank van Rey for highly dependable assistance.