The female larval fly CNS connectome: first draft

To understand how an animal thinks or moves, scientists often look at the wiring diagram of its brain, known as a connectome. This map shows every single neuron and the specific connections, or synapses, where they talk to one another. Just as two people of the same age and background might have different handwriting or distinct ways of solving a problem, even genetically identical animals can have slightly different brain wiring. To find out which parts of a brain are flexible and which are rigid, researchers need to map the brains of many individuals to see how much they vary. For a long time, creating these maps has been incredibly difficult because the structures are microscopic and the data required is massive. The fruit fly, Drosophila melanogaster, has long been a favorite subject for this kind of work because its nervous system is small enough to be manageable, yet complex enough to be interesting. While scientists have already mapped parts of the adult fly brain, the larval stage—the tiny, worm-like form that hatches from the egg—has remained largely uncharted territory, despite being a crucial time when the brain learns to control the body.

A team of researchers has now released the first complete, though still rough, wiring diagram for the entire central nervous system of a female fruit fly larva. They call this dataset "Octo." Using a powerful type of electron microscope that slices through tissue with extreme precision, they captured a three-dimensional image of the larva's entire brain and nerve cord. The resulting image is a massive block of data, free of the gaps and distortions that often plague such projects. From this raw image, the team used artificial intelligence to automatically trace the paths of neurons, locate the tiny powerhouses inside them called mitochondria, identify the supporting glial cells, and predict where the neurons connect to one another. This work provides the first full view of a larval fly's central nervous system, offering a foundation for future studies that will compare many individuals to understand how brain circuits vary and how those variations might lead to different behaviors.

The researchers started by preparing a single female larva and imaging it with enhanced focused ion beam scanning electron microscopy. This technique allowed them to see the tissue at a resolution of 8 nanometers in every direction, capturing details as small as the membranes that separate cells. The final image volume is enormous, containing billions of tiny 3D pixels that represent a physical space roughly the size of a grain of sand. To make sense of this data, the team applied a series of computer programs designed to recognize patterns. They generated three different versions of the neuron map. One version covers the entire nervous system, including the cell bodies where neurons are born and the long fibers that carry signals. Two other versions focus specifically on the dense tangle of connections in the center of the brain, known as the neuropil. By comparing these computer-generated maps against a small number of neurons that human experts had carefully checked by hand, the team found that the most comprehensive map, created using a method called Flood-Filling Networks, was the most accurate at keeping individual neurons whole without accidentally splitting them apart or merging two different neurons into one.

Beyond just tracing the wires, the team also mapped the locations of synapses, the tiny junctions where one neuron passes a signal to another. They trained a computer model on a few small, manually annotated sections of the brain where scientists had painstakingly identified hundreds of connections. Once the model learned what a synapse looked like in this specific type of image, it scanned the entire nervous system and predicted over 3.5 million potential connections. While these predictions are not yet perfect and require human verification, they provide a first draft of the entire communication network. The team also mapped the mitochondria, which are the energy-producing organelles inside cells, and the glial cells, which wrap around and support the neurons. These additional maps are important because the shape and location of mitochondria can help scientists verify if a neuron is intact, and the glial cells provide context for how the brain is organized.

The authors are clear that this release is a starting point, not a finished product. The current maps are largely unproofread, meaning that while the computer has done the heavy lifting of tracing the structures, a human has not yet checked every single line for errors. Fixing the entire dataset by hand would take decades of work, so the team hopes that their release will inspire the development of better computer tools that can use biological knowledge to correct mistakes automatically. For instance, knowing that mitochondria are usually found near the start of a connection could help a computer program decide if a detected link is real or a mistake. Despite these limitations, the dataset is a significant milestone. It is the first time the entire central nervous system of a larval fly has been reconstructed in one piece, and it is the second fully mapped brain of a first-instar female larva. By making all the raw images, the computer-generated maps, and the software tools publicly available, the researchers have handed the scientific community a powerful new resource. This allows other scientists to begin exploring how the brain of a larva controls its body, how sensory information limits movement, and how the wiring of individual flies differs from one another, paving the way for a deeper understanding of how neural circuits evolve and function.