Transcriptional noise sets fundamental limits to decoding circadian clock phase from single-cell RNA snapshots

Area of Science:

• Chronobiology
• Molecular Biology
• Genomics

Background:

• The circadian clock regulates physiological rhythms, necessitating precise phase measurement.
• Single-sample phase inference from bulk RNA offers an alternative to time-series analysis.
• The utility of single-cell RNA analysis for circadian phase remains uncertain.

Purpose of the Study:

• To assess the feasibility of single-cell circadian clock phase estimation using RNA expression.
• To determine the number and type of genes required for accurate single-cell phase inference.
• To investigate the impact of transcriptional noise on circadian phase estimation at the single-cell level.

Main Methods:

• Multiplexed single-molecule fluorescence in situ hybridization (smFISH) to quantify up to six core-clock genes in mouse fibroblasts.
• Application of a Gaussian process-based algorithm for phase inference.
• Computational simulations to predict gene requirements for accurate single-cell inference.

Main Results:

• Transcriptional noise in core-clock genes prevents reliable single-cell phase estimation, even with minimal technical dropouts.
• Simulations suggest approximately 50 low-noise, clock-like oscillatory genes are needed for accurate single-cell inference.
• Averaging smFISH counts from just three core-clock genes across ~70 cells enabled robust phase estimation.
• Single-cell resolution failed to reveal spatially heterogeneous phases in desynchronized fibroblasts; averaging did.

Conclusions:

• Circadian phase inference from RNA measurements alone requires coarse-graining (averaging) due to inherent transcriptional noise.
• Single-cell resolution is insufficient for reliable circadian phase estimation from RNA data.
• Averaging across multiple cells is essential for uncovering population-level dynamics like inter-cellular coupling.