Noise-driven morphogenesis independent of transcriptional regulatory programs

Abstract

Development is widely understood as a deterministic process driven by transcriptional programs that specify cell fate and orchestrate morphogenesis. However, this view overlooks pervasive stochasticity in gene expression, often considered an obstacle to reliable tissue patterning. Here, we introduce stochastic tuning-driven morphogenesis (STM), an alternative conceptualization of development in which noisy gene expression is not a nuisance but the primary driving force--guiding cell fates toward optimal multicellular configurations by a trial-and-error process analogous to reinforcement learning. STM operates independently of fixed transcriptional programs, instead relying on convergence of sensory information into signaling hubs, which by reinforcing random transcriptional changes, prospectively and contextually fine-tune gene expression along key developmental milestones. STM offers a fundamentally different view of development--one in which stochastic gene expression enables real-time optimization of gene expression toward multicellular objectives, implementing a self-organizing process that is inherently resistant to molecular and environmental fluctuations.