Kinetic analysis of intracellular Hoechst 33342—DNA interactions by flow cytometry: Misinterpretation of side population status?



We outline a simple approach involving instrument setup and calibration for the analysis of Hoechst dye 33342-loading in human cell lines for exploring heterogeneity in dye efflux efficiency and the status of side population (SP) A549 lung cancer cells. Dual excitation 488 nm/multiline UV (351–364 nm) flow cytometry was used to confirm ABCG2-specific inhibition of dye efflux using Fumitremorgin C. Transporter gene expression, assayed by qRT-PCR, confirmed higher expression of ABCG2 versus ABCB1, reiterated in a cloned subline. Coexpression of aldehyde dehydrogenase genes ranked as aldehyde dehydrogenase class 1A1 (ALDH1A1) > ALDH3A1 > ALDH3, relative expression of all genes was again reiterated in a cloned subline. Permeabilized cells were used to create red:violet (660:405 nm Em wavelengths) ratiometric references for mapping temporal changes in Hoechst 33342–DNA fluorescence in live cells. A live cell “kinetic SP gate” tracked progressive dye loading of the whole population and coapplication of the far red (>695 nm wavelength) fluorescing dye DRAQ7 enabled viable cell gating. Kinetic gating revealed a continuum for dye accumulation suggesting that SP enumeration is critically dependent upon the nonlinear relationship of the spectral shift with progressive dye–DNA binding and thus requires accurate definition. To this end, permeabilized cell reference samples permit reproducible instrument setup, guide gate boundaries for SP and compromised cells, and offer a simple means of comparing SP enumeration across laboratory sites/platforms. Our approach reports the dynamic range for the spectral shift, revealing noninformative staining conditions and explaining a source of variability for SP enumeration. We suggest that live cell kinetic sorting of all cells with the same dye:DNA load but with differences in efflux capacity can be used to explore drug resistance capability without prejudice. The SP phenotype should be regarded as a kinetic parameter and not a fixed characteristic—critical for functional assay design and the interpretation of heterogeneity. © 2012 International Society for Advancement of Cytometry


Profiling of cellular nongenetic heterogeneity can reveal how systems express innate variation, mount temporal responses to perturbation, and demonstrate emergent behavior in both health and disease (1). Whole population functional profiling using a “systems cytometry” approach is aimed at the recognition, analysis, and isolation of dynamic subpopulations such as normal stem cells or cancer cells coexhibiting stem-like features and multidrug resistance (MDR) that can enable post-therapy repopulation (2–4). Possible phenotypic markers for cells with stem-like properties (5) or MDR include the ATP-binding cassette (ABC) transporter superfamily members capable of facilitating the cellular efflux of xenobiotics (6, 7). The major ABC transporter genes implicated in the MDR phenotype comprise: P-glycoprotein (MDR1; ABCB1), MRP1/2 (ABCC1/2), and the breast cancer resistance protein BCRP/MXR encoded by ABCG2 (8, 9). ABCG2-linked transporter function has been used extensively in the detection and FACS isolation of pluripotential “side populations” [SPs; (10)] by virtue of the rapid energy-dependent (11) efflux of the DNA minor groove-binding UV-excitable dye Hoechst 33342 and the modified fluorescence emission spectrum in such cells (5, 10, 12–17). ABCG2 transporter substrates comprise a wide range of structurally unrelated compounds in addition to those principally used as fluorometric reagents such as rhodamine 123, Lysotracker Green™, and BODIPY-FL-dihydropyridine (18).

Although the SP phenotype is not a unique feature of stem cells (19), SP-based enrichment for cells with stem-like properties has been used with human cancer cell lines and tumors (20–27). However, it remains unclear whether the SP constitutes a distinct subpopulation, with implications for both reliable enumeration and the extent of biomarker coexpression. Here, we have focused on a kinetic approach to dye uptake in cultured human cells primarily for the drug resistance profiling. This technical note incorporates caveats, rationale, and recommendations for the wider development of SP-based analyses for cell isolation by flow cytometry—addressing a possible misinterpretation of flow cytometric analyses.

Discrepancies between gene/protein expression and efflux pump activity/turnover (18, 28) underpin the continuing need for functional cytometry-based assays that exploit fluorogenic reporters (29) and permit subpopulation enumeration and isolation (10). Cell preparation approaches, setups for flow cytometry, marker staining, and protocols for dye efflux analysis have been described previously (30), together with the technical adjustments required for cancer stem cell functional analyses (31) and recommendations for SP assay data reporting (32). However, the interpretation of SP status is complex due to the dependence of the assay on active processes and the enigmatic spectral characteristics of Hoechst 33342 dye bound to DNA in live cells (33–35). The general experience is that SP analyses are subjected to significant variability within the same cell line or between tumor biopsy samples (14). Such variations and instabilities of phenotype frustrate the exploration of biologically significant heterogeneity and the status of cells isolated by FACS (1, 32). We describe a robust kinetic approach to functional SP analysis supported by establishing a dynamic range for the assay using reference samples. Further, given the potential importance of combining loss-of-viability sensing with SP detection and immunofluorescence (32), we introduce the use of the non-UV-excitable far-red fluorescent viability sensor DRAQ7 (36).

The ABCG2 half transporter protein acts as an efflux pump for sulfate and glucuronide conjugates and for multiple anticancer agents (18, 19, 37–39). Inhibition of ABCG2-mediated transporter function by Fumitremorgin C (FTC) results in the reversal of drug resistance created by transfection of ABCG2 (40). However, the ABCG2 half transporter is yet to have a fully defined role in drug resistance in human cancers due to complex substrate specificity patterns and mutation modifying the spectrum of molecules transported (8, 37, 39–41). Thus, we have used the human A549 lung cancer cell line as a model system, which maintains a resident SP fraction and exhibits functional overexpression of ABCG2 (20, 21, 37). The A549 cell line displays evidence of reduced protein expression of ABCB1 versus ABCG2 as determined by flow cytometry (29). Aldehyde dehydrogenase class 1A1 (ALDH1A1) is highly expressed in adult stem cells, contributes to cyclophosphamide resistance, and potentially is a coexpressed marker for the SP phenotype, although there is evidence of extensive heterogeneity for functional enzyme expression in lung cancer cell lines (42). To establish the dynamic range for the SP assay in a given cell line, we have used permeabilized cells under controlled buffer conditions. We have exploited a buffer design developed for the cold preservation of human organs for transplantation, already used in cell permeabilization protocols (43) for breaching plasma membrane integrity while retaining aspects of mitochondrial function.


Cell Culture and Reagents

The A549 human lung alveolar cell carcinoma (44) (American Type Culture Collection) cells were grown in DMEM medium (Sigma-Aldrich, Dorset, UK) with 10% FCS, 1 mM glutamine, and antibiotics and incubated at 37°C in an atmosphere of 5% CO2 in air. A cloned subline of A549 was obtained by limiting dilution and expansion from a single cell to establish A549 subclone 3H2 (analyzed at approximately >25 population doublings). Hoechst 33342 (catalog number H1399; Molecular Probes, Life Technologies) was prepared as a 17.6 mM stock solution in water and added directly to culture media. FTC was obtained from Sigma (UK; catalog number F9054-250UG). DRAQ7 (catalog number SKU:DR71000; BioStatus, Shepshed, UK) was supplied as aqueous stock solution of 0.3 mM and stored at 4°C.

Flow Cytometry for Hoechst 33342 Uptake

A FACS Vantage flow cytometer was used (Becton Dickinson, Cowley, UK), equipped with a Coherent Enterprise II argon ion laser having 488 nm and multiline UV (351–364 nm) outputs (Coherent, Santa Clara, CA). CELLQuest software (Becton Dickinson Immunocytometry Systems) and filters (Omega Filters, Brattleboro, VT) were used for signal acquisition and analysis of Hoechst 33342 uptake. Optical configuration described in MIFlowCyt Supporting Information. Standard analytical flow cytometry setup was used, as configured by the manufacturer, using a jet-in-air system (70 μm nozzle) used a sheath pressure 10–11 psi. Sample differential pressure adjusted to acquire at 200–300 events per second for a sample concentration of 5 × 105 cells/mL. Time-delayed Hoechst 33342 originating signals derived from UV excitation were detected in linear mode via a 660/20 nm LP filter (nominally red emission) and via a 405/20 nm filter (nominally violet emission). For intact cell viability assessment alone, cells are incubated with DRAQ7 (3 μM) for 10 min at 37°C prior to flow cytometry using 488 nm excitation and fluorescence detection in log mode at wavelengths >695 nm (nominally far red). To establish a convenient reference setup for positioning the DRAQ7-positive fraction within a convenient/standard intensity channel, the same cell suspension was costained with the cell permeant dye DRAQ5 (20 μM × ≥100 s at 37°C; catalog number SKU:DR50200, Biostatus, Shepshed, UK) to reveal the fluorescence distributions of all nucleated cells within a sample or a preferred biological control. Correlated signals were collected for a minimum of 10,000 cells using the 488 nm forward light scatter as the master trigger signal. Signals were acquired and analyzed with CELLQuest software (Beckton-Dickinson Immunocytometry Systems).

SP and Viability Monitoring

Cells were cultured for 48–72 h prior to treatment/harvest and cells harvested at ≤70% confluence. For inhibitor studies, single-cell suspensions (5 × 105 cells/mL) are pretreated with FTC (10 μM × 1 h) in complete medium at 37°C. If used, FTC is also present during any subsequent exposure to Hoechst 33342. Typically, the protocol involves two distinct steps. (i) Adherent cells are detached using standard methods and resuspended at 5 × 105 cells/mL (1 mL) in full, prewarmed, and pregassed culture medium in individual flow tubes. Cell suspensions can be held up to 60 min at 37°C in a humidified incubator in 5% CO2 in air prior to Hoechst 33342 addition. (ii) Hoechst 33342 (prepared as a 17.6 mM stock solution in water) is added directly to culture media to give 10 μM concentration. If required, DRAQ7 is added to each tube at 3 μM DRAQ7. Incubation continued at 37°C with event acquisition typically being initiated at 5, 10, 20, 40, and 60 min for separately incubated samples.

Cell Permeabilization

Typical protocol: (i) adherent cells detached using standard methods and cells resuspended in complete culture medium at a concentration of 5 × 105 cells/mL and 1 mL samples dispensed into individual flow tubes. (ii) Tubes were centrifuged at a maximum relative centrifugal force of 150g for 10 min and pellets gently resuspended in 1 mL permeabilization buffer comprising mitochondrial respiration buffer at pH 7.1 (buffer code MiR05, see:; 0.5 mM EGTA, 3 mM MgCl2.6H2O, 60 mM K-lactobionate, 20 mM taurine, 10 mM KH2P04, 20 mM HEPES, 110 mM sucrose, and BSA 1 g/L) supplemented with 50 μg/mL of digitonin. MiR05 was prepared in advance and adjusted to pH 7.1 with fresh 5 N KOH, filter sterilized, and stored in 50 mL aliquots at −20°C in plastic vials prior to thawing and digitonin addition. (iii) Samples held at 37°C for 10 min. Completion of optimal permeabilization (digitonin concentration vs. time) for a given cell type can predetermined by the change in cellular light scatter characteristics and the simultaneous acquisition of DRAQ7 positivity. (iv) Hoechst 33342 (to provide a final concentration within the range of 0.1–20 μM) or DRAQ7 (to provide a final concentration 3 μM) added to cell suspensions and held at 37°C for 5 min prior to flow cytometric analysis. See Supporting Information for MIFlowCyt relevant protocols.

Gene Expression Analysis Using Real-Time PCR

RNA extraction: Cultures at up to 80% confluence were trypsinized and resuspended in DMEM 10% FCS spun out and washed in Hanks' Balanced Salt Solution. RNA was extracted using an RNeasy RNA extraction kit (Qiagen) according to the manufacturer's instructions. Briefly, cells were resuspended in the lysis buffer and homogenized using a QIA spin shredder column (Qiagen) before RNA was extracted using a spin column procedure. Quality and quantity of extracted RNA was analyzed using a NanoDrop Spectrophotometer (NanoDrop Technologies). Analysis of gene expression using relative quantification real-time RT PCR used cDNA generated from total RNA in 20 μL reactions using TaqMan RT reagents (Roche Applied Biosystems) as follows: 10× reverse transcription buffer, 5.5 mM MgCl2, 500 μM/dNTP, 2.5 mM random hexamers, 0.5 U/μL RNase inhibitor, 1.25 U/μL multiscribe reverse transcriptase, and RNase-free water. Two microliters of specimen RNA was added to each reaction. Reactions were performed in a thermal cycler (MJ research PTC-200) in a 45-min cycle consisting of 10 min at 25°C, 30 min at 48°C, and 5 min at 95°C. PCR primers were designed for ALDH1A1, ALDH2, ALDH3A1, ABCB1, and ABCG2 using the Roche Applied Science assay design centre and used in conjunction with the appropriate TaqMan probe (81, 26, 85, 90, and 29, respectively; Fig. 1b) taken from the Roche Universal TaqMan probe library. A commercial kit containing primers and TaqMan probes for the housekeeping genes GAPDH and peptidylprolyl isomerase A (PPIA), respectively (Roche Applied Science), was used for normalization of the reaction. Twenty-five microliters PCR reactions consisted of TaqMan 2× Universal PCR master mix (Roche Applied Science), 400 nM forward and reverse primers for the target genes, and 400 nM forward and reverse primers for GAPDH and PPIA. TaqMan probes were used at a concentration of 50 nM for all genes. Step, time, and temperature conditions for RT-PCR conditions of target genes were initial denaturation of 5 min at 95°C; denaturation step 2, 30 s at 95°C; annealing, 30 s at 59, 56, 59, 62, and 59°C for ALDH1A1, ALDH2, ALDH3A1, ABCB1, and ABCG2, respectively; elongation, 30 s at 72°C; process repeated 34 times from denaturation step 2; final elongation, 10 min × 72°C. Results were analyzed using the dCt method of real-time PCR relative quantification (previously described (45)). Cycle threshold (Ct) is the cycle number at which the fluorescence generated in the reaction well crosses the threshold value and delta Ct is the difference in Ct between the gene of interest and the endogenous control for a given sample. Statistical analysis involved a comparison of means using a paired two-tailed t-test with two-way ANOVA method, followed by a Bonferroni correction for multiple testing to allow for differing standard deviations between groups (Pvalue of ≤0.01 was used to define significance). All statistical analyses were performed using GraphPad Prism version 4.00 (GraphPad Software).

Figure 1.

A549 transporter expression and viability monitoring. Panel a: Hoechst 33342 uptake profiles in the A549. The cells were preincubated for 1 h with 10 μM FTC or DMSO alone, before incubation with 5 μM Hoechst 33342 (60 min). Data shown are representative distributions. Panel b: Details of forward and reverse primers sets for the target genes analyzed by real-time PCR. Panel c: Gene expression analyses using real-time PCR for A549 and subclone 3H2 cells. dCt values represent the difference from the reference (GAPDH) value in the number of PCR cycles, dCt values being inversely related to degree of expression, for drug transporter genes: ABCG2 and ABCB1; ALDH genes H1A1, H3, and H3A1; reference peptidylprolyl isomerase A gene: PPIA. Triplicate determinations (±SD; asterisks indicate P < 0.001). Panel d: Left panel shows the DRAQ7 staining characteristics of live A549 cells revealing a minor background of DRAQ7+ cells. Right panel shows the DRAQ7 staining characteristics of permeabilized A549 cells revealing all cells as DRAQ7+ cells. [Color figure can be viewed in the online issue which is available at]


Inhibition of Efflux in A549 Cells

SP enumeration is sensitive to culture growth phase (46). Here, culture conditions were controlled to avoid growth retardation. We initially confirmed heterogeneity for expression of ABCG2 protein in A549 cell populations by immunofluorescence (see Supporting Information). It is recognized that ABCB1 protein can facilitate Hoechst 33342 dye efflux while some ABCG2 modulators are also ABCB1 inhibitors (e.g., PSC0833) (18). First-generation MDR modulators such as verapamil (an ABCB1 inhibitor) and cyclosporine A (ABCB1/ABCG2 modulator) typically require high doses to reverse MDR and will impact upon SP profiles according to the extent of multitransporter expression. FTC can be regarded as “static” inhibitor with specificity for ABCG2 activity while “dynamic inhibitors” (e.g., PZ-39) can both inhibit activity and promote ABCG2 protein degradation (18). Functional ABCG2 activity in A549 cells was therefore confirmed using the ABCG2 inhibitor FTC to overcome the low Hoechst 33342-loading of a fraction of the cellular population (Fig. 1a). A cell cycle distribution was discernible only in A549 samples treated with FTC, suggesting that efficient intact cell DNA staining was only attained when efflux was blocked (Fig. 1a).

Transporter Expression Profiling

Overall patterns of transporter gene expression in A549 cell lines were analyzed using RT-PCR for ABCG2 and ABCB1 (47) using the primer sets shown in Figure 1b. A subcloned A549 cell line (3H2) was included to test whether expression patterns of a randomly selected clone could reiterate the parental profile. Aldehyde dehydrogenase (ALDH) encoded by ALDH1A1 is a putative functional marker of cancer stem and progenitor cells (48, 49). The study incorporated ALDH1A1 and other ALDH reference genes (ALDH3 and ALDH3A1) to assess overall changes in aldehyde dehydrogenase gene expression. GAPDH and the peptidylprolyl isomerase A gene PPIA were used as two reference genes. Normalized against a GAPDH standard, low dCt values indicated relatively high levels of ALDH expression (Fig. 1b). The profiles for the ALDH genes were similar between parent and subclone lines, showing expression ranking in the order of ALDH1A1 > ALDH3A1 > ALDH3. Moderate levels of ABCG2 expression were shown by the parent and subclone. A significantly lower level of expression of ABCB1 by the parent line, consistent with previously identified A549 cell line characteristics, was reiterated by the subclone. Further, the results were consistent with enhanced Hoechst 33342 efflux reflecting the impact of enhanced ABCG2 gene expression (50) in parental and a subcloned population.

Viability Monitoring

Although originally identified as a mitoxantrone transporter, the list of drug substrates for ABCG2 protein-mediated transport continues to expand (51) and includes topoisomerase inhibitors (e.g., anthracycline-related antibiotics and camptothecin derivatives), tyrosine kinase inhibitors (e.g., Gefitinib), and antimetabolites (e.g., methotrexate) [for review see: (18)]. Pre-exposure to such drugs is less likely to act in a competitive mode rather they may skew the Hoechst 33342 fluorescence signals of SP versus non-SP populations, due to a differential impact on cell cycle distribution, quenching of dye fluorescence by DNA binding or induction of cell death. Thus, we introduce the use of the non-UV excitable viability dye DRAQ7 to exclude cells with loss of cellular membrane integrity.

DRAQ7 is a DNA binding, far-red fluorescing viability dye chemically modified to render it membrane impermeant for intact cells. DRAQ7 has been previously used for viability sensing in melanoma cells (52) and human dendritic cells (53). DRAQ7 is nontoxic and can be used in extended culture applications and cell sorting. To establish a simple method of evaluating cell integrity and the setting of gates for use in SP analyses, live (Fig. 1d, left panel) and permeabilized (Fig. 1d, right panel) cells were exposed to DRAQ7. A background of dead cells, coincident with the staining potential of digitonin-permeabilized cells, was clearly distinguishable within nonpermeabilized control populations. The DRAQ7-DNA fluorescence signal also has the intrinsic capacity to report cell cycle status of dead cells.

Establishing a Dynamic Range for SP Gating

Early flow cytometry work described the violet (390–440 nm wavelength) to green (515–560 nm wavelength) shift in the fluorescence emissions wavelengths in chicken thymocytes (34) and human tumor cells (33) exposed to Hoechst 33342. At low concentrations or early uptake periods, the minor groove DNA binding of dye molecules shows fluorescence enhancement and a violet-biased emission spectrum. At higher dye concentrations or later uptake periods, typically used to reveal SPs, interactions between dye molecules indicative of resonant energy transfer result in a shift in emission spectrum towards the red (35). An intact plasma membrane acts to restrict Hoechst 33342 dye entry and progression of the reporting nucleus through the spectral shift. In intact cells, operation of active efflux further reduces intracellular dye availability resulting in a restrained spectral shift observed as a distinct “side population” in conventional flow cytometric bivariate plots. Emission spectrum analysis using narrow band-pass filters (centered at 400, 450, 500, 555, and 600 nm wavelengths) reveals that dye–DNA dissociation during the incubation of mammalian cells in dye-free medium results in preferential loss of red versus violet fluorescence. The reverse spectral shift is accelerated when overall dye efflux is enhanced 10-fold by specific overexpression of ABCG2 (19, 54). This kinetic relationship can be exploited to identify moribund cells which show an “instantaneous” and distinct red-biased Hoechst 33342–DNA emission signature (55) due to ready access of dye to DNA binding sites. We have used this approach with permeabilized cells to establish the dynamic range for SP analyses under controlled buffer conditions in addition to viability gating (see above).

We initially applied a simple gating strategy for the identification of intact A549 cells by light scatter in combination with exclusion of the viability dye DRAQ7 used at 3 μM (Figs. 2a and 2b). Gating of intact cells (R1 + R5) permits exclusion of dead cells on the basis of 488 nm (or alternatively 633 nm) excited DRAQ7-DNA fluorescence during Hoechst 33342 uptake analysis. The Hoechst 33342 dose dependency for a violet to red spectral shift was determined using digitonin-permeabilized cells (Figs. 2c and 2d) permitting the extraction of the ratio values (Fig. 2e) that correspond to equilibrium conditions for different dye:DNA values according to the concentration and type of cells used (see Supporting Information Fig. S2 for typical bivariate plots). Intact cells showed a variation in the relative red and violet fluorescence signatures for Hoechst 33342 at an exemplar uptake period of 40 min, defining a region for SP (e.g., Fig. 2f; R2) that can be matched with a region occupied by a permeabilized reference sample (e.g., Fig. 2c). This approach overcomes the anomalies observed in the assay (46) when dye:cell DNA ratios are not controlled, given the sensitivity of violet versus red fluorescence to intracellular dye concentration, or when the impact of dye-depletion by nonviable cell binding is not assessed.

Figure 2.

A549 SP analysis. Panel a and b: Light scatter defining region R1 (a) and DRAQ7 exclusion defined by region R5 (b) provide a gating strategy for the identification of intact, live cells (R1+R5) based on normal light scatter and the exclusion of DRAQ7 or the spectral analysis of Hoechst 33342 fluorescence. Panel c–e: Spectral characteristics of digitonin-permeabilized cells exposed to 0.5 (panel c) or 10 μM (panel d) Hoechst 33342 × 10 min showing typical setup for bivariate distributions and the corresponding region settings for regions R2, R3, and R4 representing the potentially restricted, normal, or maximal loading of live cells. Panel c shows a typical dose dependency of Hoechst 33342 spectral shift characteristics for digitonin-permeabilized cells given by ratio of red/violet fluorescence (ratio 1 = channel 400). Panel f shows a typical SP (R2; 13%), normal (R3; 77%), and maximally loading cells (R4; 2%) for 10 μM Hoechst 33342 × 40 min.

Dye titration of permeabilized cells permits reproducible instrument setup for SP analyses (Fig. 3a) and provides a reference plot for the spectral shift sensitivity of the system. Comparison of the red:violet ratios for permeabilized and live cells (Fig. 3b) shows that the standard deviation for the whole live cell population extends into the noninformative range for dye concentration reporting. Comparing non-SP cells with matched permeabilized cell reference samples for ≥1 μM dye (Fig. 3c) shows a fivefold range of equivalent dye loadings with a mode equivalent to a 2.5 μM dye loading—this range comprises both dye uptake and DNA content variation. Ratio values below that achieved by ≤0.5 μM dye are not sensitive to dye concentration. In principle, the cutoff is based on the physical principles limiting the attainment of significant dye–dye interactions. However, below the cut-off violet fluorescence emission values are sensitive indicators of uptake as they are less influenced by dye–dye interactions. Comparing a gated SP population with a 0.5 μM dye permeabilized cell reference sample (Fig. 3d; Kolmogorov Smirnov statistical analysis, D = 0.37, P < 0.001) reveals that ∼20% SP cells have a dye loading equivalent of <0.5 μM but ≥0.1 μM. Thus, for a 40 min Hoechst 33342 exposure period, there is a >fivefold difference in the accumulation of dye between SP region and non-SP region designated cells. Overall, the range for dye–DNA interactions in the whole population is of the order of 50-fold representing significant levels of target protection expressed by some cells.

Figure 3.

Dynamic range for SP assay. Panel a: Reproducible Hoechst 33342 concentration dependency of normalized relative red versus violet Hoechst 33342 fluorescence in permeabilized cells (dotted line shows polynomial fit for fluorescence ratio (y) and dye conc. (x); y = −0.0011 × 2 + 0.0882x + 0.261; R2 = 0.9976). Panel b: Relative red versus violet Hoechst 33342 fluorescence in permeabilized cells compared to intact cells (40-min exposure). Panel c: Distribution of red/violet fluorescence values for intact non-SP cell population (R3; 10 μM Hoechst 33342 × 40 min; shaded distribution) versus permeabilized cell reference samples (nonshaded distributions). Panel d: Distribution of violet fluorescence values for intact SP cell population (R2; 10 μM Hoechst 33342 × 40 min; shaded distribution) versus permeabilized cell reference samples (nonshaded distributions).

Kinetic SP Gating for Whole Population Profiling

Methodological variations exist for reported SP analyses and often use an approach of stain, chill, wash, and hold in buffer on ice (46). Such an approach is useful for the FACS isolation of rare SP cells that may require extended instrument run-times. However, an early study has shown ongoing efflux in SP-like cells has an initial T½ of significantly <30 min for the loss of red fluorescence, allocating non-SP cells to an SP fraction, highlighting the need to control the potential for on-going dye wash-out (54).

A kinetic SP gate (R2 combined with either DRAQ7 negative status and/or normal light scatter) was established, through which all cells must pass during the course of Hoechst 33342 uptake period (Fig. 4a) or potentially reverse during a chase period in dye-free medium (54). Exit from a kinetic SP gate boundary indicates acquisition of sufficient dye molecules on DNA to invoke dye–dye interactions. The gate boundaries can be established by red:violet ratios, a permeabilized control sample, or the functional impact of an efflux inhibitor. It is clear that as the dye exposure period extends then the frequency of cells in the SP-gated region continues to decline while any background of compromised cells remains constant. The results suggest that there is a continuum for expression of SP-like characteristics with respect to dye loading period. The kinetic gate does allow for the definition of SP membership at a given time point (e.g., 40 min or SPt40; Fig. 3b) without implying irreducible membership.

Figure 4.

Kinetic gating enrichment of SPs. Panel a: Kinetics of progression of A549 cells through an SP defined region (R2) during exposure to 10 μM Hoechst 33342. Flow cytometric data (•, SP cells; ○, non-SP cells; Δ, cells with compromised membrane integrity showing maximal staining) were gated for intact cells. Trendline fits (x = time; y = % gated cells): non-SP data, y = 29.254ln(x) − 28.567, R2 = 0.9541; SP data, y = −32.06ln(x) + 133.29, R2 = 0.9686). Panel b: Frequency distribution SP% determined at 40-min dye uptake (i.e., SPt40; see results text) of A549 cell culture for 34 independent determinations (8.1 ± 7.2, mean ± SD).

Repetitive SPt40 analyses yield a wide distribution of SP40 values (Fig. 4b) with a range of 2–28% and a mode of ∼4% (mean 8.1 ± 7.2 SD) suggesting that even under controlled assay conditions, biological variation can impact upon SP enumeration. Upon initial isolation and culture expansion (∼20 population doublings), the A549 3H2 subclone displayed a high SPt40 fraction (32%). Upon further culture (over 20 passages), the subclone SPt40 fraction decreased and stabilized (9.4 ± 2.7%; n = 5) to within the normal range for SP maintenance at which it displayed parent-like gene expression patterns as shown in Figure 1c. A lineage slowly reiterating the parental distribution argues against the existence of self-sustaining, stable subpopulations that can vary their relative frequency through differences in proliferation.

Caveats and Recommendations

  • 1SP-based measurements cannot be used to imply the operation of a specific transporter or to define the transporter selectivity of efflux inhibitors, without ancillary profiling of cells for ABC transporter expression. The application of an ABCG2-specific inhibitor such as FTC can be used to negate Hoechst 33342 efflux and provide a useful demarcation of cells expressing ABCG2-mediated efflux with respect to a chosen boundary for the SP characteristic (19).
  • 2Permeabilized cell reference samples permit reproducible instrument setup for kinetic analyses, indicate assay performance for a given session, and offer simple reference plots for comparing the efficiency of SP enumeration across laboratory sites and platforms.
  • 3Permeabilized cells can be used to match cell and Hoechst 33342 concentrations to optimize dynamic range for different optical filter combinations and provide a ratiometric cut-off for SP gate boundaries.
  • 4The spread of bifluorescence values within the SP region, as defined by permeabilized cells, reflects the expected degree of the heterogeneity within the whole population for DNA access and not efflux. Thus, data spread within the SP region for intact cells also incorporates aspects of this heterogeneity—potentially misconstrued as differences in efflux efficiency or degree of transporter expression.
  • 5Different permeabilization buffers and protocols, optimized for a given cell type, can be used to establish convenient SP reference samples. Given that dye efflux is energy dependent, we have opted for a “respiration” buffer system to aid compatibility with intact tissue samples.
  • 6Loss of membrane integrity results in cells showing unimpeded Hoechst 33342 uptake and a red:violet fluorescence signature equivalent to permeabilized cells stained at the active dye concentration. Application of a viability dye DRAQ7 without UV-coexcitation can be used to exclude moribund cells from efflux analyses by multicolor gating.
  • 7The frequency of cells occupying a SP region at a given time point (e.g., SPt40) is highly dependent upon active dye concentration and cell density and provides only a “snap-shot” of a kinetic process.
  • 8A generic step-by-step procedure is given in Supporting Information Table S1.


SP analysis uses a useful but “arbitrary” break point in the nonlinear ratiometric Hoechst 33342 dye–DNA fluorescence signature. We functionally define the SP region in bivariate fluorescence plots as the attainment of time-dependent dye loading without quenching. All cells can enter this region during dye loading but exit as quench effects dominate. Thus, the SP characteristic is a kinetic parameter and not a fixed phenotype—critical for the interpretation of heterogeneity and when comparing SP-analyses across different laboratories. Assay conditions (46) and biological variation affect dye loading rates making SP enumeration both subjective and nonrobust. We reiterate the importance of kinetic analyses in dye efflux studies (54, 56) and show that kinetic gating can be used to time profile the continuum of efflux competence of a whole viable cell population, reporting the degree of SP enrichment with respect to time. Cells can be sorted on the basis of the same dye:DNA loading but with different levels of SP enrichment based on a time parameter, removing the potential prejudice of differential exposure to the toxic and phototoxic action of the Hoechst 33342 dye (19, 57) in FACS applications. Single-cell efflux profiling can also underpin modeling studies on the origin and maintenance of heterogeneity. We conclude that the continued technical efforts to control and refine cytometric analyses and thereby constrain data interpretation are vital for the exploration of transporter phenotype stability, heterogeneity, and coexpression with other properties related to MDR or stemness.


PJS and RJE are directors of Biostatus Ltd., suppliers of the viability probe DRAQ7. Exemplar list mode files for the data presented are available from the corresponding author.