Let’s take a minute to talk about connectomics. No, not genomics. No, not metabolomics. Not any of the other -omics, but connectomics. It’s a new-ish field that the computational neuroscience geek in all of us can love.
By way of introduction, the “connectome” is the “network of elements and connections forming the human brain” (according to Sporns et al, 2005). Let’s forget the part about human brain, and (for the purposes of this post) say that a connectome is the set of all the neural connections in a nervous system. Connectomics is the subfield of neuroscience that attempts to understand the structure and function of nervous systems by studying these connections as a whole. The goal of this discipline is to determine the anatomy of the nervous system to a very fine scale (on the order of individual cells), and then relate this structural information to the functioning of the nervous system. “Solving” a connectome is achieved when the connections between all the neurons in a nervous system are mapped. This has been done for C. Elegans, a nematode who has only 302 neurons, by the use of electron microscopy – this work is summarized in White et al., 1986. Since then, we’ve expanded our goals, with the Human Connectome Project aiming to solve the connectome of the human brain. Let’s step back for a second, though, and ask: why do we want to know all of this?
Since the elucidation of the electrical properties of neurons (which, by the way, started with the squid giant axon – you can read one of my older posts about that topic here ), neuroscientists have been interested in the information that neurons carry. A nervous system can be thought of as an organ that processes information to tell the rest of the body what to do. Some stimulus might impinge upon a sensory organ (let’s say, for example, that one sees a car speeding towards one’s self,) which causes a cascade of electrical activity through the nervous system, eventually leading to such diverse effects as the movement of one’s muscles to carry one’s self out of the path of the car, the emotional distress that occurs when one is almost killed by some jerk who isn’t paying attention to the road, and the realization that one has just gotten very lucky. Later on, one might tell the story of this near miss to her friends over dinner, exhibiting the ability of the nervous system not only to process, but to store information for later use. Further demonstrating this ability, one might learn not to walk in the middle of the road in the future. The ability of animals to exhibit behavior, to move, feel, and learn, is all due (according to the dogma of neuroscience) to the processing of information by cells in the nervous system.
Now, it’s relatively routine to study how a single neuron processes information. To sum it up very briefly, information enters a “typical” neuron in the form of electrical impulses on that neuron’s dendrites. The information flows through the neuron (as an easy-to-get analogy, imagine electronic information flowing through the wire) and then a decision is made: at any given time, if a neuron is electrically excited enough, it can discharge an action potential. An action potential is a burst of electrical activity that will travel through the neuron’s axon to affect the activity of other neurons that the axon makes synapses with. Thus, one can imagine a general flow of information through a neuron from the dendrites, through the cell body, and out the axon. A nerve cell is diagrammed below, with the dendrites and axon clearly labeled, showing the flow of electrical impulses through the neuron. Such neurons are linked together to form functional circuits that accomplish complex tasks such as recognizing objects, coordinating movement, and recognizing when food tastes good or bad (to name but a few.)
Think of it this way: each neuron works like a tiny computer processor. At any given time, it’s integrating all of the electrical signals coming to it, and deciding whether or not to fire an action potential. Nervous systems can process information because they have many such relatively simple processors connected together to process that information. (Keep in mind that this is a very simplified, and thus necessarily inaccurate description of the nervous system. It will have to do for now, however, and gets across those aspects of the function of neurons that are most relevant to the problem of the connectome well enough.) Thus, to understand the function of a nervous system, one only has to understand the functioning of each of its neurons. This turns out to be incredibly difficult, and at best we only achieve approximations of this goal. I’ll come back to this later on.
To study how a single neuron processes information is relatively easy, if we’re selective about which neurons we study. For example, we could record electrical activity from the optic nerve while we expose the eyes of the animal we’re studying to light. In this way, we would see the way that different visual signals are encoded by neurons in the optic nerve. In fact, this has been done countless times in studies on the visual system in cats, frogs, ferrets, and many other species. Using this method of electrical recording, we can determine how neurons are functionally connected to each other, and how they respond to various inputs. However, to record electrical information from a neuron requires that one physically place an electrode into the brain, and also requires that one focus on a single neuron or a small subset of neurons at a time. To figure out how an entire brain functions in terms of the interaction of millions of individual neurons would be all but impossible using these electrophysiological methods. In addition, this method can never be employed on humans, as it is considered unethical to put electrodes into peoples brains without an urgent medical need (of course!)
Another now-classical way of learning about the connections between neurons in a nervous system is through tracer studies. In these studies, neurons in one part of the brain are dyed in some way. Then, other parts of the brain (or the whole brain) can be examined to see if they have dye in them. If they do, it’s concluded that the neurons make some sort of connection between the part of the brain where the dye was injected and the part of the brain where the dye was later seen. This has many of the same downfalls as electrophysiological methods of figuring out neural circuits. For one, it can only be done in a small group of neurons in any preparation, and so the connections in the nervous system must be mapped out piecemeal, a few at a time. In addition, it is often difficult to tell what route an axon might take from the cell body to its destination, even if it is clear where each of these points are.
The difficulty that these methods have in resolving the microscopic structure of the nervous system beg for a faster, more flexible technique. Even the reconstruction of the relatively simple nervous system of C. elegans, done using images from an electron microscope, had to be done largely by hand. These processes are labor- and time-intensive, and do not lend themselves well to the reconstruction of nervous systems that may have millions or billions of neurons.
Enter the field of computational neuroscience. Computational neuroscientists study nervous systems in terms of their information processing capabilities. Standing at the junction of computer science and neuroscience, they have both the tools and the impetus to understand the details of the connectomes of whichever organisms they study.
An approach that has been taken in humans involves using the technique of diffusion tensor imaging, and MRI technique that can determine the direction that axons run in in an intact brain. For example, the following image (by Thomas Schultz) shows a DTI-derived image of the connections that run through the midline of a living human brain:
Such images are of great potential use in studying brain lesions, doing studies on brain function, clinical diagnosis, and whole-brain level analysis of neural circuits. However, they lack the resolution needed to map individual synapses, thus falling short (for the time being) of being able to comprehensively map the connections between neurons in a brain. For this, we have to go to microscopy techniques that involve looking directly at neural tissue. These can only been done in animals, because it is presently illegal to harvest brain tissue from humans for experimentals purposes (again, a no-brainer.)
By now, you’re wondering when I’ll mention a cephalopod. After all, this is a blog about cephalopods. You have every right to expect that I’ll mention squids, octopodes, or nautiloids at least once in each post. Never fear!
Last week I came across a poster presentation write-up on Biomed Central called “Charting out the octopus connectome at submicron resolution using the knife-edge scanning microscope.” As you might imagine, I was tickled. A research team with members from Texas, Naple, Michigan, Illinois, California, and Seoul (including Graziano Fiorito, notable for his research on observational learning in the octopus) is working on reconstructing the octopus connectome using a mostly-automated 3D microscope called the knife-edge scanning microscope (KESM). This microscope takes a block of tissue and slices it, taking a picture of each slice of the tissue as it is cut. Then, a computer program can create a high-resolution 3D image of the tissue. From this, the computer can (and this is the tricky part) automatically trace the paths that nerve cells take through the tissue, and – this being the goal of this research – reconstruct a detailed network showing the morphology of each neuron in the tissue. For examples of the resulting images, you can see this gallery from the brain networks laboratory at Texas A&M.
Why the octopus? Well, in an introduction that makes the comparative neuroanatomist in me jump for joy, the authors suggest that because “the neural architecture of this cephalopod mollusk differs markedly from that of any vertebrate… [investigating] the difference and simlarities between the neural architecture – or connectome – of the octopus and mammals, such as the mouse, may lead to deep insights into the computational principles underlying animal cognition.” In their concluding remarks, the authors note that they “expect that this pilot study and the more detailed investigations to follow will allow fruitful comparisons of the neural circuitries of individual octopuses with different ecological life histories, as well as of animals that have been exposed to a variety of neurodegenerative insults… In sum, this approach should contribute greatly to our understanding of the computational architecture of invertebrates and ultimately provide insights into the differences between invertebrate and vertebrate cognitive capabilities.”
I’m intriguied by this article, but also a little dissappointed. Mostly, I’m dissappointed that a more complete study isn’t out yet! I’ll be watching these guys from now on, and I’ll cover any other publications they put out on the topic. Hurrah for octopus connectomics!
In closing, I want to mention that a complete neuroanatomical picture of a nervous system does not actually explain its computational properties. To understand how nervous systems process information, we need to know the physiology of each cell and the biochemistry of the interactions, a topic that is probably more complex than even the very fine-grained study of neuroanatomy represented by the studies I’ve mentioned here. In terms of our understanding of nervous systems, however, connectomics offers and opportunity to study the relationship between the cellular structure of the nervous system and its overall capabilities – a relationship whose description has been one of the goals of neuroscience practically since its inception.
Thanks for reading!
Sporns, O., Tononi, G., & Kötter, R. (2005). The Human Connectome: A Structural Description of the Human Brain PLoS Computational Biology, 1 (4) DOI: 10.1371/journal.pcbi.0010042
Yoonsuck Choe, Louise C Abbott, Giovanna Ponte, John Keyser, Jaerock Kwon, David Mayerich, Daniel Miller, Donghyeop Han, Anna Maria Grimaldi, Graziano Fiorito, David B Edelman, & Jeffrey L McKinstry (2010). Charting out the octopus connectome at submicron resolution using the knife-edge scanning microscope BMC Neuroscience, 11 (Supplement 1), 136-137 : 10.1186/1471-2202-11-S1-P136
MAYERICH, D., ABBOTT, L., & McCORMICK, B. (2008). Knife-edge scanning microscopy for imaging and reconstruction of three-dimensional anatomical structures of the mouse brain Journal of Microscopy, 231 (1), 134-143 DOI: 10.1111/j.1365-2818.2008.02024.x
White, J., Southgate, E., Thomson, J., & Brenner, S. (1986). The Structure of the Nervous System of the Nematode Caenorhabditis elegans Philosophical Transactions of the Royal Society B: Biological Sciences, 314 (1165), 1-340 DOI: 10.1098/rstb.1986.0056
MORI, S., & ZHANG, J. (2006). Principles of Diffusion Tensor Imaging and Its Applications to Basic Neuroscience Research Neuron, 51 (5), 527-539 DOI: 10.1016/j.neuron.2006.08.012