How many neurons comprise a neural column

Scientists from the Max Planck Society create a detailed anatomical map of the neuronal networks in the brain

The new research results form the basis of computer-aided studies of the healthy and diseased brain.

In three closely related studies, researchers at the Max Planck Institute in Florida have presented one of the most comprehensive analyzes to date of the detailed structure of an essential part of the neuronal networks of the brain: the “cortical column” of the cerebral cortex. The three studies made it to the cover of the October issue of the journal Cerebral Cortex (Volume 20, Issue 10). In the same issue appeared a special comment from Edward G. Jones of the University of California and Pasko Rakic ​​of the Medical School at Yale University.

The extremely accurate analysis includes an anatomical overview of the exact dimensions as well as the number, distribution and type of nerve cells (neurons) in a normal cortical column. In addition, the sensory inputs emanating from the thalamus are precisely measured in a column. The quantitative and spatial measurement of the brain's neural networks is one of the most important tasks in neuroscience and one of the indispensable steps on the way to an understanding of how the brain works. The results of the three studies enable the creation of complex computer models of the neural networks in cortical columns and represent an important intermediate step on the way to mapping the neural networks of the entire brain.

"We are researching the neuronal networks involved in processing sensory stimuli in the cerebral cortex of animals in order to understand how the brain works," reports Hanno-Sebastian Meyer, a researcher at the Department of Digital Neuroanatomy at the Max Planck Florida Institute and the lead author from two of the studies. “We selected the networks that are functionally linked in the rodent brain to the whiskers used for touching. Ultimately, with our data, we will be able to create a computer model of this system. The results of the work that has now been published represent the first important step. "

While the importance of a detailed reconstruction of the neural networks in columns was often not recognized or was not possible with conventional methods in earlier studies, these investigations provide an important quantitative view of these archetypal structures of the cerebral cortex. Although it was performed on animal models, the reconstruction of these neural networks opens up a multitude of new perspectives. The researchers hope to gain a better understanding of the changes in the network architecture that are associated with sensory and cognitive anomalies.

A quantitative approach

Large parts of the cerebral cortex, which play a central role e.g. in sensory perception, memory functions and language, consist of six horizontal layers of nerve cells. The nerve cells are functionally and structurally organized in vertical columns. When sensory organs are stimulated (e.g. when deflecting a whisker), neurons in the parts of the cerebral cortex responsible for processing sensory stimuli (in the case of whiskers, this is the "somatosensory" cerebral cortex) first become active in the same column. These pillars can be directly assigned to the sensory organs "one to one" - in rodents there is a "responsible" pillar for each beard hair. It is therefore assumed that the neural networks in these pillars play a major role in the primary processing of the data received from the sensory organs. The thalamus, a collection of nerve cells near the middle of the brain, is an intermediate station for the sensory stimuli (sight, hearing and touch): It forwards the information to the corresponding pillars in the sensory areas of the cerebral cortex.

Perhaps the most promising aspect of the new investigations, according to the commentary on the work, is the quantitative method chosen by the researchers. This procedure involved the careful and elaborate localization of all neurons in identified columns of the somatosensory cortex. The number (approx. 19,000) and three-dimensional distribution of the nerve cells in a column could thus be determined.

The researchers were surprised to find such a large number of neurons in the individual columns, according to Meyer. Previously it was assumed that the number was closer to around 10,000 - which corresponds to half of the actual value. "It is obvious that precise knowledge of the number and distribution of neurons is required if one wants to create accurate computer-based models."

Mapping of neural networks

In order to quantitatively record the number and distribution of all neurons in complete columns, the researchers labeled the neurons with fluorescent chemicals that could be visualized using sophisticated microscopy techniques.

The location of the individual neurons was initially determined by hand, with special software being used to visualize and mark the neurons.

“Within two years, we manually marked all neurons in three columns and the adjacent areas and determined the number and distribution in the columns,” said Meyer. “During this process, we obtained many terabytes of image data from these parts of the brain and marked more than 100,000 neurons. This manually collected data will be the starting point for the future development of algorithms for automatic localization. "

The result of this image acquisition, localization and analysis is a vast pool of data on the structure of the cortical columns. In combination with measurements of the activity of individual neurons, this allows scientists to estimate the potential output of a normal cortical column with unprecedented accuracy: around 4400 signals are emitted by the nerve cells in the various layers within 100 milliseconds after a whisker is deflected.

The work also included the quantification of the contacts originating from the thalamus, which the different types of nerve cells receive in a column. These results provide the anatomical basis for functional measurements at the level of individual nerve cells.

“By combining functional data with our new knowledge of anatomy, we can estimate and better understand the input and output signals in a cortical column when a beard is touched,” said Meyer. "Here, too, the accuracy of our quantitative approach is of crucial importance for the creation of a reliable computer model."

In their future research, Meyer and his colleagues will move on to developing methods that will ultimately allow a map of the entire rodent brain to be created. "We want to capture the structure of the other pillars in the sensory cortex," said Meyer. "Then we will analyze, among other things, the brains of mice suffering from Alzheimer's to determine what is happening with the neural networks in the early stages of this disease happens. ”The analysis of the cellular structure of complete brains is still a long way off, but researchers are on the way.

"The mapping of neural networks of complete brain areas and ultimately the entire brain at the single cell level is such a large project that we have to think about how it can be done on a large scale, similar to what has been successfully demonstrated in the human genome," he said.

The Digital Neuroanatomy Group at the Max Planck Institute in Florida is headed by Bert Sakmann, MD, Ph.D. who received the Nobel Prize for Medicine in 1991 together with the physicist Erwin Neher. The group focuses on research into the functional anatomy of neural networks in the cerebral cortex. They represent the neural basis of simple behaviors, such as making decisions. This research requires the use of extensive, high-resolution microscopy techniques in order to reconstruct the individual structures, positions and synaptic connections of different neuron types in large volumes. Eventually, this will reveal the parts of the network that control sensory-triggered behavior and lead to new discoveries about learning in the brain.