The human brain is estimated to contain 100 billion neurons (the number 1 followed by eleven zeros). Because a typical neuron forms ~1,000 synaptic connections to other neurons, the total number of synapses in the brain is estimated to be 100 trillion (the number 1 followed by 14 zeros). The thin projections from neurons that form connections with each other (axons and dendrites) can be thought of as the biological “wiring” of the brain.
Neuroscientists already know that brain neurons can and do form specific rather than random connections with each other to generate the observed wiring diagram of the brain. However, the precise patterns of such non-random connections, how the patterns are formed, and how these patterns underlie the brain’s extraordinary information processing capacity are important questions that Cold Spring Harbor Laboratory theoretical neuroscientist Dmitri Chklovskii and others are exploring. An article published in this week’s issue of PLoS Biology (March 1, 2005) describes Chklovskii’s discovery of strongly preferred patterns of connectivity or scaffolds within the wiring diagram of the rat brain. The patterns are likely to correspond to modules that play an important role in brain function not only in rats, but also in humans.
Chklovskii and his colleagues use statistical analysis and mathematical modeling–coupled with in vivo, experimental observations–to search for recurrent, non-random patterns of local connectivity within the vast thickets of brain wiring diagrams. Finding such patterns would be strong evidence for the presence of functional modules (for example, “local cortical circuits”) that process information. The researchers recently uncovered evidence of such functional modules by using two complementary approaches.
In the first study–published in December–they chose the nematode worm C. elegans as a relatively simple model system. Studies by others had determined that this organism has 302 neurons, and had mapped which neurons connect with which. However, those studies did not characterize non-random patterns of connectivity in a rigorous way.
When Chklovskii and his colleagues considered all 13 possible patterns of connectivity that can occur among three neurons (one such “triplet” pattern being “neuron A connects to B, B connects to C, and A connects to C”), they found that three particular patterns, including the aforementioned one, stood out as appearing far more frequently in the C. elegans wiring diagram than they would by chance. They also discovered that some triplet patterns were less common than predicted by chance. Taking the analysis a step further, Chklovskii found that among all 199 possible patterns of connectivity that can occur among four neurons, one particular pattern stood out in C. elegans as appearing more frequently than it would by chance.
Significantly, Chklovskii considered whether the frequent connectivity patterns or “motifs” they discovered might be accounted for by previously known principles of neurobiology. They found no such explanation for the existence of the motifs, indicating that further analysis of the motifs may reveal important information about nervous system structure and function.
Because it was based purely on anatomical data collected by electron microscopy, Chklovskii’s C. elegans study did not include telling information about the strengths of connections between neurons. Therefore, to extend his findings into the physiological realm, Chklovskii collaborated with researchers at Brandeis University on the study published this week in PLoS Biology. The Brandeis group had previously collected one of the largest electrophysiological data sets of its kind ever recorded: measurements of the connectivity of some 3,000 individual neurons in the rat visual cortex.
Chklovskii realized that the Brandeis data could be used to explore his ideas concerning functional modules in the brain. He and his colleagues detected some of the very same non-random patterns of connectivity in the rat brain as they had observed in C. elegans. More importantly, they found that most connections formed by neurons in the rat visual cortex are weak, and that the stronger connections (~17% of all connections) account for as much as half of the total synaptic strength of a particular network. In part because more strongly connected neurons fire more reproducibly, Chklovskii proposes that strong cortical synapses–with particular connectivities–act as a network “scaffold” that is likely to generate reproducible patterns of activity and play an important role in brain function. “Local brain circuits can therefore be viewed as a ‘skeleton’ of strong connections in a sea of weaker ones,” says Chklovskii.
