Multiplexed automated imaging assay for cardiotoxic compounds using the ImageXpress Pico system

Introduction

Oksana Sirenko, PhD | Sr. Research Scientist | Molecular Devices

There is an increased need for expanding variety and complexity of cell-based assays for biologic research and drug discovery. Stem cell-derived cells and tissues become an increasingly attractive alternative to traditional in vitro and in vivo testing in pharmaceutical drug development and toxicological safety assessment. In this study, we used human induced pluripotent stem cell (iPSC)-derived cardiomyocytes to develop functional and morphological readouts for testing effects of different compounds in a multi-parametric assay format.

We performed automated cell imaging and analysis of iPSC-derived cardiac cells with the ImageXpress^® Pico Automated Cell Imaging System to simultaneously determine calcium oscillation frequency, cell viability, cytoskeletal integrity, apoptosis, and mitochondrial function. Effects on cardiomyocyte beating frequency were characterized by measurements of calcium oscillations. Multiplexed assessment of different readouts provides additional insight into the mechanisms of action of various compounds. The methods were characterized using a set of known cardio-active drugs and selected cardiotoxic compounds.

Stem cell-derived cell models for compound screening

Stem cell-derived cardiomyocytes, as well as liver cells and neurons, provide very useful models for compound testing and toxicity assessment. Cardiotoxicity remains one of the main reasons for drug attrition during clinical trials. In addition, a significant percentage of cardiovascular diseases are reportedly due to environmental exposures. Accordingly, assay development for in vitro screening for potential toxic effects is an important area of investigation.

Methods

Instrument

Cell-based assays were performed using the ImageXpress Pico system in combination with CellReporterXpress^™ Image Acquisition and Analysis Software. The imager provides four fluorescence channels, transmitted light, and time-lapse capability to enable automatic monitoring of complex biological responses in live cells.

Environmental control and time-lapse monitoring

The ImageXpress Pico system is equipped with an environmental control chamber which enables control and monitoring of temperature, CO₂ and O₂ content, and humidity. In combination with time-lapse imaging, the system is an efficient tool for performing live cell experiments under normal or hypoxia conditions.

Cell culture

Human iPSC-derived cardiomyocytes and the appropriate media were purchased from Cellular Dynamics International, Fujifilm Co. (CDI). Cells were plated into 384-well black clear bottom plates at a density of 10,000 cells per well and cultured as recommended by protocols from CDI. Treatment with compounds was performed for 24 hours.

Cell staining

To visualize Ca²⁺ oscillations, cells were loaded with calcium dye from the EarlyTox^™ Cardiotoxicity Kit (Molecular Devices). To assess phenotypic changes, live cells were stained using a mixture of three dyes: viability dye Calcein AM (1 μM), mitochondria potential dye MitoTracker Orange (0.2 μM), and nuclear dye Hoechst (2 μM) (all from Life Technologies). For visualizing the actin cytoskeleton, cells were fixed with 4% formaldehyde (Sigma-Aldrich) and stained with AlexaFluor 488 (AF488) labeled phalloidin stain.

Assessing compound effects on calcium oscillations in cardiomyocytes

iPSC-derived cardiomyocytes are a very attractive in vitro model. They form a synchronously beating monolayer that can be used to reliably reproduce drug-associated cardio-physiologic phenotypes using a fast, kinetic fluorescence assay that monitors changes in intracellular calcium oscillations (Grimm et al. 2016; Sirenko et al. 2013). In this work, we adapted the calcium oscillation assay for the ImageXpress Pico system which combines time-lapse imaging with environmental control.

Cardiac cells spontaneously contracting and physiological movements can be observed using transmitted light imaging. Fluorescent imaging of intracellular Ca²⁺ fluxes was performed on the ImageXpress Pico system using environmental control, with images set to be acquired at 0.5 second intervals. After loading cells with calcium dyes, fluctuations of fluorescence intensity consistent with the mechanical contractions were observed. Figure 1A shows images taken by time-lapse acquisition of cardiac cells stained with calcium dye (EarlyTox Cardiotoxicity Kit).

Images were analyzed using the Cell Count module to detect cells and measure the average fluorescence intensity and average cell area. Calcium oscillations were visualized by plotting fluorescence intensity over time.

For evaluation of cardiotoxic effects, cardiomyocytes were treated with compounds, typically for 24 hours. Figures 1B and 1C present various examples of calcium oscillation traces generated after addition of tested compounds. Figure 1B represents a screenshot from an experiment testing six representative compounds in a dose-dependent manner (half-log changes in concentrations). Figure 1C shows enlarged representative traces for control, doxorubicin, and flecainide acetate conditions. Note the inhibition of calcium flux as an effect of doxorubicin and prolongation of flux with flecainide acetate treatment.

Figure 1. Measuring Ca²⁺ oscillations for iPSC cardiomyocytes. To measure Ca²⁺ oscillations, iPSC cardiomyocytes were stained with EarlyTox Cardiotoxicity kit then imaged over time in the FITC channel using the ImageXpress Pico system. (A) The panel above presents a series of images set to be acquired at 0.5 sec intervals. (B) Traces of calcium oscillations shown for increasing concentrations of indicated compounds. (C) Traces comparing control (0.1% DMSO, yellow), doxorubicin (1 mM, blue), or flecainide acetate (1 mM, purple) treatments.

Changes in the frequency of calcium oscillation, amplitude, and peak shapes were observed as a result of compound treatments. The number of peaks were counted manually. Peak counts were evaluated in duplicates and concentration-dependencies were plotted using SoftMax^® Pro Software (Figure 2). EC₅₀ values for inhibition of oscillation frequency were calculated from a 4-parameter curve fit.

Figure 2. Number of peaks (counted manually) plotted against compound concentration using a 4-parametric curve fit. EC₅₀ values were calculated using SoftMax^® Pro Software. Dose-dependent decreases in peak count shown for cisapride (purple), haloperidol (red), staurosporine (blue), imatinib (yellow), doxorubicin (dark green), flecainide acetate (green) treatments.

Assessing compound effects on cell viability and morphology

While the evaluation of changes in beating profiles is important for detection of functional effects, imaging also provides an essential complementary assay for monitoring morphological changes and cytotoxicity effects of compounds. Imaging and analysis provide important tools for characterization of multiple readouts including cell viability, characterization of cell shape, cell adhesion and spreading, cytoskeleton integrity, and mitochondria membrane potential.

We used a live cell staining protocol that enables a onestep addition of a mix of three dyes which eliminates the need for fixing cells or performing repeated wash steps. Calcein AM was used to identify viable, metabolically active cells and stain for whole-cell morphological features. Cell-permeable nuclear dye Hoechst was utilized to measure total cell counts and assess nuclear shape. MitoTracker Orange was used to detect cells with intact mitochondria and measure impact of tested compounds on mitochondria potential. Cells were stained after 24 hour incubation with compounds.

Images were taken with 10X or 20X objective, in three colors: DAPI for Hoechst nuclear stain, FITC for Calcein AM, and TRITC for MitoTracker Orange. A representative composite image is shown in Figure 3A. Images were analyzed using cell scoring or multi-wavelength cell scoring analysis modules. The analysis algorithm defines first the nuclei, then scores cells as positive or negative depending on the intensity of Calcein AM or MitoTracker Orange stains. The decrease in the number of positive cells, cell area, or staining intensities indicates toxicity effects. Figure 3B shows images of control and damaged cells, and the analysis masks for Calcein AM. EC₅₀ values for different measurement values were derived from concentration-dependency plots and are shown in Table 1.

Figure 3. Image analysis of cardiomyocytes. (A) Image of cardiomyocytes (20X magnification) stained with viability dye Calcein AM, Mitotracker Orange dye for detection of mitochondria with intact membrane potential, and Hoechst nuclear dye (all from Thermo Fisher Scientific, 0.5 μM, 0.2 μM, and 0.5 μM, respectively). (B) Images and the analysis masks for multi-parametric analysis. iPSC-derived cardiomyocytes were treated with compounds for 24 hours and then stained with a mixture of Hoechst 33342, Calcein AM, and MitoTracker Orange CMTMRos. Images and analysis masks were compared for control cells and cells treated with 0.1 μM staurosporine and 10 μM doxorubicin. Cells were imaged with DAPI, FITC, and TRITC filters using a 10X Plan Fluor objective. The images show nuclei (in blue), Calcein AM stain (green), and mitochondria (orange). Images were analyzed using the Cell Scoring analysis module optimized for quantitation of Calcein AM positive cells or MitoTracker Orange positive cells. The analysis masks: light green—positive nuclei, red—negative nuclei, green—actin cytoskeleton. EC₅₀ values are shown in Table 1.

*Numbers of live cells (positive for Calcein AM stain), also the total area covered by live cells, and average intensities of Calcein AM stain were measured using image analysis.
**Blank cells indicate no effect at highest concentration tested (50 μM).

Table 1. Multi-parametric evaluation of compound effects on cardiac cells. EC₅₀ values for inhibition of frequency of calcium oscillation are shown in comparison with EC₅₀ values for cytotoxicity readouts related to cell viability, spreading, and intensity of Calcein AM signal.

Multi-parametric analysis of phenotypic changes allows detection of toxicity and also provides information about the mechanism of toxicity. Notably, most compounds demonstrated cardiac specific toxicity, with EC₅₀ concentrations for inhibition of beat rate at least ^½ log or lower than corresponding concentrations for cytotoxic effects as measured by a decrease in number of Calcein AM positive cells. In contrast, the cytotoxic effect of doxorubicin was observed at a similar concentration as the inhibition of the beating rate.

Conclusion

Our results demonstrate how a variety of assays can be utilized for quantitative screening of chemical effects in iPSC cardiomyocytes and enable rapid and cost-efficient multidimensional biological profiling.