Breimann, Laura et al. (2021) International Worm Meeting "Precise quantification of mRNAs across all C. elegans embryonic stages through a microscopy and machine learning-based approach"

Breimann, Laura, Bahry, Ella, Preibisch, Stephan, Harrington, Kyle I S, Epstein, Leo, Lionnet, Timothee, Ercan, Sevinc, Kolyvanov, Klim

[International Worm Meeting, 2021]

Precise quantification of mRNA transcripts in space and time throughout embryogenesis is essential for understanding gene regulation, a process critical for embryogenesis in all animals, including C. elegans. We developed an imaging approach using 3D widefield microscopy-based single-molecule RNA fluorescence in situ hybridization (smFISH) to quantify mRNA transcripts. To count individual single-molecule mRNA spots, we developed RS-FISH, a fast 3D spot detection method that we implemented in Fiji that combines radial symmetry and RANSAC outlier removal. To assign each fixed, imaged C. elegans embryo to its developmental stage, we used advanced machine learning-based image classification that relies on the concept of auto-encoders. Currently, we are applying our methods to understand the role of condensins in chromosome compaction and transcription regulation. In C. elegans, an X-specific condensin binds to and represses X chromosomes in XX hermaphrodites by 2-fold for dosage compensation. In our study, we want to understand condensin DC's effect on transcript numbers and dynamics in single embryos across development. We obtained thousands of smFISH images for a set of condensin DC-regulated and control genes and extracted mature and nascent RNA counts in 3D, which we use to determine transcription burst characteristics throughout embryonic development. The distribution of total transcripts in wild-type and condensin DC-depleted embryos shows that single genes on the X chromosome are downregulated ~2-fold. Our machine learning approach to separate embryo images by development stage allowed us to observe the timing of condensin DC-mediated transcription repression, which occurs from the 100-cell stage on. RS-FISH is freely available as a Fiji plugin, and details for installation can be found at https://github.com/PreibischLab/RadialSymmetryLocalization and described at https://doi.org/10.1101/2021.03.09.434205