In a computer model of a fly, adding 100 cells of the right types after the loss of 2000 neurons restored displacement to 60% of normal, while 100 cells of other types restored it to only 6,7%
In a computer model of a fly, adding 100 cells of the right types after the loss of 2000 neurons restored displacement to 60% of normal, while 100 cells of other types restored it to only 6,7%
On October 27, Pardes Habib, Michael Petr and Ruslan Rust posted a preprint on bioRxiv that has not been peer reviewed. Using a map of connections in the nervous system of an adult male fruit fly, they modeled 165 122 neurons and tested which added cells restored movement in the simulated fly.
Habib is based at the Stanford Stroke Center, and Rust leads a brain cell therapy laboratory at the University of Southern California. Stroke kills neurons and causes lasting movement impairments. Treating this damage by replacing cells requires knowing what determines recovery: the number of cells, their types or their connections. These factors are difficult to disentangle experimentally, whereas a fly model allows researchers to specify exactly which cells are removed and restored.
Each model neuron accumulates signals through connections in the map and fires when it reaches a threshold, following the rules of the 2024 fly brain model developed by Philip Shiu and colleagues. Movement in the visual field activates direction-selective cells, wing motor neurons control turning and speed, and a simplified fly moves across a plane for ten seconds while exposed to wind gusts. Displacement is the distance from the starting point; results are expressed as percentages of displacement in the healthy model.
Removing 1000 neurons around a single point in the optic lobe left displacement at 99,8% of normal; removing them around a point in the posterior central brain reduced it to 43,2%. The second lesion removed 37 descending neurons that carry commands from the brain to the nerve cord (the equivalent of the spinal cord), including the giant fiber involved in escape takeoff. Removing only these 37 neurons left displacement at 57,4%, while removing 37 other cells in the same region left it at 101,0%.
Removing 2000 neurons around the second point left displacement at 6,5% of normal. The authors restored cells at the sites of the lost neurons, each with its original connections, selecting descending neurons and visual cells for their roles in movement control. Adding 100 such cells, including all 79 lost descending neurons, restored displacement to 62,5% of normal. Adding 100 randomly selected cells restored it to 10,1%, and adding 1000 randomly selected cells restored it to 48,5%; on average, a random selection of 1000 includes half of those 79 neurons.
With the same 100 selected cells, wing motor neuron activity reached only 40,9% of normal, while displacement reached 62,5%. The model calculates forward speed from the average of wing motor neuron activity and the escape takeoff response to an approaching object; this response reached 90,5% of normal.
The authors then inferred the connections of the new cells from their types, assuming that a transplanted neuron would connect in the same way as other neurons of its type. Each new cell was assigned either the type of the lost cell (the correct type) or a different type, and its connections were copied from a surviving cell of the assigned type. Adding 100 cells of the correct types restored displacement to 60,0%. On average, the connections for 89 of these cells were copied from their mirror counterparts on the opposite side of the brain, whose connections closely resembled those that had been lost. Adding 100 central brain neurons of other types, with connections copied from those types, yielded 6,7%, almost the same as without replacement. When all 2000 cells were replaced, the correct types restored displacement to 87,6%, other types to 9,3%, and the correct types with random connections to 14,3%. “Replacing as many cells as had been lost was insufficient when their types or connection partners differed from those in the original circuit,” the authors write.
The authors propose testing these findings in living flies by genetically ablating descending neurons in the posterior central brain, or by reversibly silencing cells and then reactivating them. Reactivation would approximate replacement with the original connections preserved. In the worm C. elegans, researchers predicted which neurons were needed for movement from a map of connections and confirmed the predictions by removing individual cells.