Structural Evolution of Neural Networks during Zebrafish Early Development

Abstract

The spatial architecture of neural networks underpins nervous system function, yet dynamic in vivo evidence for their early establishment process remains scarce. Here, using long-term confocal live-cell imaging of transgenic zebrafish, we dissected neural network development across the brain, yolk sac, and spinal cord at single-cell resolution. We reveal that repeated growth cone bifurcation is the fundamental cellular mechanism for network node formation. Despite distinct developmental origins, networks in all three regions exhibit highly conserved dynamics, strictly following Strahler hierarchical ordering and Hack's power-law scaling. Quantitative topological analysis shows that as development progresses, node numbers and clustering coefficients increase, while network density decreases. This reflects a structural optimization strategy balancing high functional efficiency with low wiring cost. These findings suggest that physical and geometric constraints, combined with simple local rules, drive the self-organization of complex neural networks, offering valuable principles for artificial network design.