Fig. 6: CCDC141 non-additively interacts with TTN and IGF1R to modify cardiomyocyte morphology.

a, Human induced pluripotent stem cell (iPSC)-derived cardiomyocytes with and without hypertrophic cardiomyopathy (carrying an MYH7-R403Q mutation) were transfected with scramble siRNA or siRNAs specifically targeting single (CCDC141, IGF1R, and TTN) or combined (CCDC141-IGF1R and CCDC141-TTN) genetic loci prioritized by lo-siRF. b, Gene-silenced cardiomyocytes were bifurcated into two focused streams of large and small cells using a spiral microfluidic device (cell focusing mechanism illustrated in Extended Data Fig. 6) to allow high-resolution single cell imaging. c, Workflow of the image analysis process. Time-lapse image sequences of single cells passing through the top and bottom microchannel outlets (panel b) were fed into a customized MATLAB-based program that extracts cell size/shape features via a sequential process of bright field background correction, cell boundary detection, cell tracking and stuck cell removal, cell feature extraction, data quality control and postprocessing, and morphological feature analysis. Extracted single cell features for each gene-silencing condition were compared with their scrambled control values to validate the potential role of epistasis in the genetic regulation of cardiomyocyte hypertrophy (d-j). d, Violin plots of cell diameters of unaffected (blue) and MYH7-R403Q variant (red) cardiomyocytes. Solid and dashed lines in box plots represent median and mean values, respectively. Asterisks indicate significant difference (***p < 1E-36, Wilcoxon signed rank test). e, Gene-silencing efficiency in unaffected (blue, n = 5 to 9) and MYH7-R403Q variant (red, n = 3) cells based on RT-qPCR analysis (details in Methods). Error bars indicate standard deviations. f, Percent change (relative size difference) in median cell diameter of gene-silenced cardiomyocytes relative to scramble control values indicates that CCDC141 interacts with IGF1R to rescue cardiomyocyte hypertrophy. Relative size differences were averaged across data from two to four independent batches of cells. Error bars indicate standard deviations computed on 1000 bootstrap samples of these batches with the following sample size: n = 13147 (TTN), 19460 (IGF1R), 45304 (CCDC141), 19979 (CCDC141-TTN), and 26135 (CCDC141-IGF1R) for unaffected cells and n = 22134 (TTN), 33801 (IGF1R), 21158 (CCDC141), 39515 (CCDC141-TTN), and 52049 (CCDC141-IGF1R) for MYH7-R403Q variant cells. Asterisks indicate significant difference between gene-silencing and scramble control conditions based on the maximum p-values of Wilcoxon signed rank test across all batches of cells (*p < 0.05, **p < 0.001, and ***p < 1E-4). g, CCDC141 non-additively interacts with IGF1R (left) and TTN (right) to modify boundary and texture features of unaffected (blue) and MYH7-R403Q variant (red) cells. Cell boundary waveness and texture irregularity were measured by the roundness error (h, top) and normalized peak number (h, bottom), respectively. i, Representative single-cell images overlapped with detected cell boundaries (red lines) show that a higher roundness error indicates increased irregularity of the cell boundary. j, Representative single-cell images with detected peaks (blue plus signs) of the brightfield intensity distribution enclosed within the cell boundaries (red lines) indicate a varying level of cell textural irregularity. Scale bars: 10 μm. Detailed statistical information of cell morphology measurements and non-additivity analysis for the studied gene pairs can be found in Extended Data 8.