High-Throughput Phenotyping Toolkit for Characterizing Cellular Models of Hypertrophic Cardiomyopathy In Vitro

Figures

• 

  Figure A1. Cell-type control for α-actinin staining. Comparing signal intensities of α-actinin between fibroblasts or any cell type that does not express the cardiac marker (cell-type control) with cardiomyocytes enables determination of positive signal threshold required for high-content imaging. Scale bar = 100 μm.

• 

  Figure A2. Controls for BNP signal thresholding. In order to categorize signal intensities of cardiomyocytes expressing BNP, endothelin-1 treatment can be used to determine high-intensity threshold as it maximises BNP expression relative to untreated cells, whereas bosentan treatment has the opposite effect (defining low/nonexistent signal). Scale bar = 100 μm.

• 

  Figure A3. Parameters for machine learning step in the sarcomeric disarray analysis sequence. The PhenoLOGIC™ training relies on manual assignment of cardiomyocytes to two different categories (A, organised; B, disarrayed). The parameters of algorithm using the training set provided by the user indicate goodness of fit factor (>1) as well as the texture/ morphological properties evaluated by the software that differ the most between the two categories.

• 

  Figure 1. Workflow showing toolkit for phenotyping hypertrophic cardiomyopathy (HCM) in cellular models. Cardiomyocytes can be obtained from different sources (biopsies or cell lines) and require dissociation into a single cell suspension prior to phenotyping. Evaluation of in vitro hallmarks of HCM is done via: (1) Flow volumetry to investigate hypertrophy; (2) high-content imaging to assess multinucleation, brain natriuretic peptide (BNP) hypertrophic marker expression, and sarcomeric disarray; (3) Seahorse assay and qPCR to assess mitochondrial respiration and content, respectively.

• 

  Figure 2. Protocol for evaluating hypertrophy by flow cytometry. (A–C) Forward (FSC) and side (SSC) scatter values of calibration beads of known sizes are measured, and debris and clumps are excluded by gating. (D,E) Each bead has a distinctive FSC value that is used to plot a calibration curve (FSC vs. size). (F–H) Cardiomyocyte samples are analysed and gated similarly to calibration beads to measure FSC values, which are used to determine cell volume.

• 

  Figure 3. Multinucleation Analysis Sequence by high-content imaging. Representative micrographs of image analysis of human pluripotent stem cell-derived cardiomyocytes (hPSC-CMs). After inputting the image (A–D), the software excludes debris and cell clumps (E–H) and identifies different degrees of multinucleation based on the distance between nuclei in cardiomyocytes (I–L). Scale bar = 100 μm.

• 

  Figure 4. Cardiomyocyte purity analysis sequence. Representative micrographs of image analysis of hPSC-CMs. After inputting the image (A), the software excludes debris and cell clumps (B) and uses the cell mask staining to define cell cytoplasm (C). The α-actinin signal intensity is quantified in the cytoplasm and a positive signal threshold is determined by comparing with cell-type control (e.g., fibroblasts) (D), enabling the distinction of cardiomyocytes (E) from non-cardiomyocytes (F), required to calculate purity. Scale bar = 100 μm.

• 

  Figure 5. Hypertrophic marker BNP analysis sequence. Representative micrographs of image analysis of hPSC-CMs. After inputting the image (A–C), the software excludes debris and cell clumps (D,E) and uses the cardiac troponin T (cTnT) staining to define cell cytoplasm (F). The cTnT signal intensity is quantified and a positive signal threshold is determined by comparing with cell-type control enabling exclusion of non-cardiomyocytes from the analysis (G). A perinuclear ring is defined in cardiomyocytes (H) and BNP intensity is quantified in this region. Comparison of BNP signal intensity to that of ET1- or BOS-treated cardiomyocytes enables distinction between high vs. medium BNP-expressing cells (I–L). BNP—brain natriuretic peptide; cTnT—cardiac troponin T; scale bar = 100 μm.

• 

  Figure 6. Sarcomeric disarray analysis sequence. Representative micrographs of image analysis of hPSC-CMs. After inputting the image (A–C), the software uses mathematical transformation algorithms to sharpen sarcomeric signal (D), followed by exclusion of debris and cell clumps (E,F). The cell mask channel is used to define cell cytoplasm (G) and the α-actinin signal intensity is quantified and compared against that of a cell-type control to exclude non-cardiomyocytes from the analysis (H). Morphologic and texture-based information is extracted from the sharpened α-actinin signal displayed by cardiomyocytes (I), followed by a machine-learning procedure where the user manually selects cardiomyocytes displaying organised vs. disarrayed sarcomeres (J). The software then applies this algorithm to the rest of the images being analysed to distinguish organised vs. disarrayed cardiomyocytes (K,L). Scale bar = 100 μm, except for L where it is 20 μm.

• 

  Figure 7. Evaluation of mitochondrial respiration using Seahorse platform. (A) Protocol summary for mitostress test using the Seahorse assay™. (B) Representative micrographs detailing normalisation by automated quantification of cell nuclei in Cellavista™ plate imager ((i) Hoechst staining of nuclei; (ii) quantification of individual nuclei; (iii) overlay). (C) Sample graph for oxygen consumption rate (OCR) over time following the Seahorse mitostress kit, highlighting the different mitochondrial respiration parameters.

• 

  Figure 8. Example of application of phenotyping toolkit to HCM cellular model. (A,B) Hypertrophy analysis by flow volumetry indicating increased median volume in HCM cardiomyocytes relative to healthy counterparts. (C–F) High-content imaging evaluation of HCM cardiomyocytes showing increased proportion of multinucleation, expression of hypertrophic marker BNP, and sarcomeric disarray relative to healthy controls. (G–J) Assessment of mitochondrial respiration in Seahorse™ assay revealing higher energy production in HCM cardiomyocytes in comparison with healthy ones, showing (K) similar mitochondrial DNA content (determined by ratiometric mito/nuclear DNA qPCR analysis). Data represent mean +/- SEM of five independent experiments comparing R453C-β-myosin heavy-chain hPSC-CMs (HCM) to isogenic wild-type control (healthy), with 100,000–250,000 cells analysed per sample. Statistical analysis was performed by unpaired parametric Student’s t test between HCM and Healthy conditions, with the exception of Panel E, where unpaired one-way ANOVA test was used with Dunnett’s post hoc test for correction of multiple comparisons relative to Healthy condition (n. s., non-significant; *P < 0.05; **P <…