Decoding the Human Brain, With Help From a Fly
By NICHOLAS WADE
Published: December 13, 2010
Taiwanese researchers have managed to bar code some 16,000 of the 100,000 neurons in a fruit flyâ€™s brain and to reconstruct the brainâ€™s wiring map.
In terms similar to those that define computers, the team describes the general architecture of the flyâ€™s brain as composed of 41 local processing units, 58 tracts that link the units to other parts of the brain, and six hubs.
Biologists see this atlas of the fly brain as a first step toward understanding the human brain. Six of the chemicals that transmit messages between neurons are the same in both species. And the general structure â€” two hemispheres with copious cross-links â€” is also similar.
â€śI think this is the beginning of a new world,â€ť said Ralph Greenspan, a neurobiologist at the University of California, San Diego. Biologists should now be able to match the fruit flyâ€™s well-studied behaviors to the brain circuits established by the new atlas, he said.
The atlas is maintained on a supercomputer in Taiwan which fly biologists around the world can query. They can also add to the atlas by uploading their own images of fruit fly neurons. â€śSo I think this will really accelerate progress,â€ť said Josh Dubnau, a neurobiologist at the Cold Spring Harbor Laboratory on Long Island.
The Taiwan team is led by Ann-Shyn Chiang, who has been working on the project for the last decade. He has assembled a group of 40 people, who include computer programmers and engineers, working on a budget of about $1 million a year.
The basis of the atlas is a technique for visualizing the three-dimensional structure of individual neurons, including the cellâ€™s nucleus, its long axon, and the little branches, or dendrites, with which it makes contact with other neurons.
The complex structure of a neuron can be made apparent with a green fluorescent protein modeled on one used by jellyfish. The gene for the protein is inserted into the fruit flyâ€™s genome, along with another gene that represses it. Dr. Chiang developed a technique for lifting the repression on the gene in just one neuron at a time. When the gene is expressed, the green fluorescent protein reaches every part of the neuron, defining its structure in exquisite detail.
He also invented a remarkable solvent for making the Drosophila brain transparent. This is essential if the glowing green neuron is to be imaged precisely. The solvent is so effective that if a researcher fails to keep an eye on the dissected brain as it lies on a microscope slide, the brain will simply disappear when the solvent is added, Dr. Dubnau said.
Each flyâ€™s brain is a different size and shape, so Dr. Chiangâ€™s team had to define average dimensions for the female and the male brain, creating a virtual brain with standard dimensions. They then developed algorithms for recasting the 3D image of each neuron so as to bring it into register with the standard brain. This means that the 16,000 neuron images, each taken from a different fly, can all be compared.
Each neuron is then given a bar code with the coordinates of where its cell nucleus lies within the standard Drosophila brain, as well as information about which other parts of the brain the neuron connects to, and which kind of chemical transmitter it uses.
A major setback occurred partway through the project when Dr. Chiang found he could gather data five times as fast if he recorded the neuron images in a different way. â€śPainfully,â€ť he said in an e-mail, â€śwe had to throw all the old data away,â€ť even though 3,000 neurons had already been imaged.
The neuron bar codes are numerical data that can be manipulated by computer. With 16,000 images in hand, Dr. Chiangâ€™s team was able to analyze the general architecture of the female fruit flyâ€™s brain. The basic element, which they call a local processing unit, is a group of neurons with connecting interneurons that do not extend beyond the group. Tracts of longer-range neurons connect the local processing units with one another.
The local processing units correspond with the known anatomical regions of the fly brain. They are the same in all flies, and handle specific tasks like taste or movement.
The fly brain turns out to be â€śa hybrid system of grid computing and a supercomputer,â€ť Dr. Chiang said. â€śIt tells us how a complex brain is put together and operates. Given the growing evidence for conservation in genetic programs underlying brain development and function, the human brain is likely to consist of similar basic operation units.â€ť
The only nervous system so far explored in greater detail is that of the C. elegans roundworm, another laboratory organism. But the little wormâ€™s system has only 302 neurons and perhaps does not fully deserve to be called a brain. The fly brain, with its 100,000 neurons, may prove a better starting point for understanding the human brain, which has an estimated 100 billion neurons, each with about 1,000 synapses.
â€śThe beauty of this paper is in the completeness of what he did; itâ€™s in the foresight it took to develop over a decade or more a whole suite of new methods to tackle a problem they saw as fundamental, â€ť Dr. Dubnau said, referring to the Chiang teamâ€™s work. Dr. Chiangâ€™s report is published in the latest issue of Current Biology.
â€śYesterday I almost fell out of my chair,â€ť said Olaf Sporns, who designs computer models of neural circuits at Indiana University. The matrix showing the interconnectivity of the fly brain in Dr. Chiangâ€™s article struck Dr. Sporns as amazingly similar to the matrix he had constructed recently for the human cortex.
The construction of the fly and mammalian brains seems to follow the same â€śsmall worldâ€ť principle, that of high local clustering of neurons, together with long-range connections. â€śSo thereâ€™s a commonality here, and I think that has to do with the fact that these systems have to accomplish similar goals,â€ť Dr. Sporns said.
â€śResearchers may now be able to pinpoint how information flows through the fly brain network to accomplish certain outcomes,â€ť he said.
Dr. Chiang said he will continue to build his atlas until all 100,000 fly brain neurons have been imaged. He said he does not at present plan to map the synapses, the precise connections that one neuron makes with others.
Dr. Greenspan, however, said it should be possible in principle to map synapses by splitting in two the gene for the green fluorescent protein used to delineate the neurons. The neurons could be made to export the half-proteins to their synapses, and when the two halves fused, they would glow green and let the synapse be scanned and mapped.
With a full wiring diagram of the fly brainâ€™s neurons and all their synaptic connections, researchers could test their ideas about how information flowed in the brain, and even compute the output that should follow a given input.
â€śItâ€™s not out of the question that if we had a complete cellular map and a good database, that we could create virtual organisms,â€ť Dr. Sporns said.
Click to view image: 'flypic'
|Liveleak on Facebook|