Whole-brain neural network analysis (connectomics) using cell lineage-based neuron-labeling method

The brain is a computing machine that receives input signals from sensory neurons, calculates best responses to changing environments, and sends output signals to motor muscles. How such computation is materialized remains largely unknown. Understanding the entire wiring network of neural connections in the brain, which is recently called the connectomics (connection + omics), should provide indispensable insights on this problem. To resolve the circuit diagram from the tangled thickets of neural fibers, only a small subset of neurons should be visualized at one time. Previous studies visualized such selective cells by injecting dyes or by detecting specific molecules or gene expression patterns using antibodies and expression driver strains. These approaches were unfortunately not efficient enough for identifying all the brain cells in a comprehensive and systematic manner. Neurons are generated by neural stem cells. The entire neural population can therefore be divided into a finite number of families - or clones - of the cells that are the progeny of each single stem cell. The central brain of the fruit fly Drosophila melanogaster consists of about 15,000 neurons per side and is made by utmost 100 stem cells. By genetically labeling one of such stem cells and tracing the projection patterns of its progeny in the adult brain, we were able to identify the neural projections of almost all the clonal cell groups. To visualize these neural projections, we made serial optical sections of the fly brain using laser confocal microscopy. Because of its relatively small size (0.6-mm wide and less than 0.3-mm thick), the entire fly brain can be imaged using high-resolution objectives with n.a. 1.2. Neuronal fibers are visualized by ectopically expressed cytoplasmic and membrane-bound fluorescent proteins, and the output synaptic sites are visualized with ectopically expressed tag proteins that are fused with the proteins associated with synaptic vesicles. In addition, density of all the synapses was visualized using an antibody against synaptic proteins. Signal distributions of the latter data were put into the template using 3D non-linear elastic morphing. The data of other channels are morphed with the same formula. By doing so, neural projection data obtained from different brain samples can directly be compared in the 3D space. Neurons generated by each stem cell turned out to form lineage-specific sets of projections that arborize in distinct regions of the brain, arguably serving specific aspects of brain functions. We named such clonally associated projection units the clonal units. Different parts of the brain were formed by distinct sets of overlapping clonal units. By tracing the projections made by each clonal unit, we were able to establish a comprehensive connection diagram between different brain regions. Microscopic analysis of neuronal fibers thus leads to the understanding of the entire brain neural network.