Figure 1. Panels A, B, and C from Figure 4 in the paper “Linking Macroscopic with Microscopic Neuroanatomy using Synthetic Neuronal Populations” published October 23, 2014 in PLoS Computational Biology). (A) Visualization of the dentate gyrus model highlighting 1,000 synthetic dendritic trees (dark purple structure scattered through model). (B) Rendering of the complete morphologies for all granule cells in a 20 µm transverse slice from the center of the model dentate gyrus. (C) Rendering of 48 granule cells from the crest of the slice in (B).

Most signal integration and transformations happening in your brain are happening in the neuropil. The neuropil is the tangle of extremely small processes connecting nerve cells together. Axons from nerve cell bodies touch dendrites of receiving neurons through an intervening tiny gap called the synapse through which chemicals from the transmitting cell address the receiving cell.

It’s been analytically surmised for many decades now that the shape, branching, and other properties of neuron processes influence signal processing and, although technically difficult, a number of experiments have shown this to be true. These experiments are performed on just one or a few neurons and show that, for example, a branch in an axon may cause a signal (an action potential) to slow down, or even stop, at the branch. The signal may continue down one branch but may cease to exist the other branch (this is called filtering).

If morphology has such large affects on signal processing in a single cell, what would be the effects of hundreds of thousands or millions of cells with diverse morphologies on the signal processing in a particular organ of the brain? The research team behind the article “Linking Macroscopic with Microscopic Neuroanatomy using Synthetic Neuronal Populations” (published October 23, 2014 in PLoS Computational Biology) doesn’t address this question directly but carried out work that lays a foundation for helping to answer it.

In this paper, the authors set out to build an anatomically and morphologically realistic model of a well studied organ of the brain known as the dentate gyrus located in the hippocampus. This area of our brains is particularly interesting for its central roll in forming memories and helping us to navigate through our environment. To study details of signal processing in the dentate gyrus we must take into account the significant variability in neuron morphology across this structure. Currently we can only hope to study details in signal processing across hundreds of thousands or millions of cells using simulation tools. Therefore, this team set out to construct an anatomically and morphologically realistic model of the dentate gyrus.

Note: The authors have posted the Matlab source code used to create this synthetic dentate gyrus at “Generation of granule cell dendritic morphology (Schneider et al. 2014)” record in the SenseLab ModelDB repository at . The paper doesn’t state the level of computer power necessary to generate the model but the research team did use a high performance computer cluster so it’s probably safe to say “a lot.” Please keep this in mind if you decide to download the code and play.

Rip a donut in half and squish that half circle of a donut on a countertop so that the arch almost completely collapses and each end is splayed out a bit. Now you have something resembling a rat’s dentate gyrus. A rat’s dentate gyrus is estimated to contain about 1.2 million granule cells (see Figure 1A above). Using mathematical and computational methods that capture both its overall anatomical shape (the squished half donut) and the morphology of each of over 1 million cells, the research team generated 1.19 million granular cells packed at the appropriate density into a three-dimensionally appropriate structure. They used regional statistical variation to capture experimentally observed variability in neuron processes across the dentate gyrus. The result is an impressive anatomical model that can be used to study the effect of anatomical and morphological heterogeneity on signal processing in the dentate gyrus.

Figure 1. Octave 3.8 graphical interface version displaying neuron to neuron connectivity matrices for whisker related somatosensory cortex (for more about these data see my July 9, 2011 post or the paper “Laminar Analysis of Excitatory Local Circuits in Vibrissal Motor and Sensory Cortical Areas” published January 4, 2011 in PLoS Biology).

More than three years ago on this blog I introduced Octave while writing about brain circuitry data. Octave has come a long way since then. After downloading and a simple setup on your computer you’ll notice two Octave applications. One is the traditional client and the other is the new graphical user interface version, which is slated to be the standard in version 4.

If you’d like to try running the code that displays brain circuitry data like in Figure 1 above, go to the “Laminar analysis of excitatory circuits in vibrissal motor and sensory cortex (Hooks et al. 2011)” record in the SenseLab ModelDB repository and download mhconmatvalues20100928_octave.m from the model files. Place this file in a location you’ll remember.

After starting up Octave-gui you will see a File Browser area in a left area of the application’s window. Navigate to your copy of mhconmatvalues20100928_octave.m and double click on it so that it loads into the editor (in the area to the right). The result should look similar to Figure 1 above without the six graphic display windows. Find the arrowhead (or right-pointing blue triangle) in the editor’s toolbar and click on it. This runs the file displayed in the editor. The six figures defined in the file should display.

Octave is maturing into a very attractive freely available and open source alternative to Matlab. Soon on this blog we’ll look at how easy or difficult it may be to run code written for Matlab in Octave.

This week an article was published that shows when a specific set of brain cells is activated, the animal eats whether they’re already fed or not (see the video above; “The Inhibitory Circuit Architecture of the Lateral Hypothalamus Orchestrates Feeding” published September 27, 2013 in Science). The target location of these neurons, the hypothalamus, has been known to be important in the control of eating behaviors but the precise brain circuitry driving the three “F”s – fighting, feeding and sex – has not been shown. The new research describes what looks to be a key piece of brain circuitry involved in eating behaviors.

The scientists genetically manipulated neurons in a structure known as the bed nucleus of the stria terminalis so that, when light was shined on them through fiber optics, they became electrically active and released their inhibitory chemicals (neurotransmitters) onto target neurons in the lateral hypothalamus. This decreased (inhibited) the activity of excitatory neurons in the lateral hypothalamus that contained a specific protein known as the vesicular glutamate transporter-2 protein. The result was a dramatic increase in feeding behavior.

In other words, excitatory neurons in the lateral hypothalamus seem to suppress feeding behavior. Decrease the activity of those neurons and feeding behavior is turned on. They appear to act as an on-off switch for eating. The potential therapeutic importance of these neurons is clear, especially in the United States with the highest rate of obesity in the world. However, recent public exposure of processed food industry practices presents troubling ethical dilemmas.

Last week I attended a lecture by the journalist Michael Moss where he talked about his new book “Salt Sugar Fat: How the Food Giants Hooked Us.” Moss states that much of the problem with obesity in the United States is due to the abundance of cheap, calorie-rich, processed food. He goes further and presents evidence that the companies producing processed foods use scientific research data to design the food they market to be maximally attractive to consumers. Makes sense. A company wants to sell product. Best if it sells a lot of product over and over again. The dilemma arrises when we contemplate when a line is crossed from making extraordinarily attractive food to creating additive substances.

It would be interesting to feed the mouse in the video above sugary cereal to see if this same “switch” was turned on. The research team may have begun describing part of the very circuit activated when a person experiences pleasure, termed bliss in the industry, that companies target when developing a product. It’s interesting that funding for the study included drug and alcohol abuse agencies. This is a complex, fascinating, and very important topic. I highly recommend reading “Salt Sugar Fat: How the Food Giants Hooked Us.”