Machine Learning-Assisted Decoding of Temporal Transcriptional Dynamics via Fluorescent Timer
Investigating the temporal dynamics of gene expression is crucial for understanding gene regulation across various biological processes. Using the Fluorescent Timer protein (Timer), the Timer-of-cell-kinetics-and-activity (Tocky) system enables analysis of transcriptional dynamics at the single-cell level. However, the complexity of Timer data has limited its broader application. Here, we introduce an integrative approach combining molecular biology and machine learning to elucidate Foxp3 transcriptional dynamics through flow cytometric Timer analysis. We have developed a Convolutional Neural Networks (ConvNet) approach that incorporates image conversion and Gradient-weighted Class Activation Mapping (Grad-CAM) for class-specific feature identification at the single-cell level. Biologically, we developed a novel CRISPR mutant of Foxp3-Tocky lacking the Conserved Non-coding Sequence 2 (CNS2), which has successfully elucidated CNS2-dependent Foxp3 transcription dynamics, revealing novel roles of CNS2 in regulating Foxp3 transcription frequency under specific conditions. Furthermore, generating new data from WT Foxp3 Tocky mice at various ages, the Grad-CAM methods successfully revealed distinct dynamics of Foxp3 expression from neonatal to aged mice, highlighting prominent thymus-like features of neonatal splenic Foxp3+ T cells. In conclusion, our study uncovers previously unrecognised Foxp3 transcriptional dynamics, establishing a proof-of-concept for integrating CRISPR, Tocky, and machine learning methods as advanced techniques to understand transcriptional dynamics in vivo.