Junk DNA: "Listen to your junk man - he's singing"

"Listen to your junk man - he's singing ... All dressed up in satin, walking past the alley..." - Bruce Springsteen, New York Serenade

Junk DNA is looking mighty fine lately. Only a few years ago, the non-coding regions of DNA that make up over 95% of the genome were looked upon as the uninteresting desert wastelands between the regions of DNA involved in protein synthesis. How times have changed!

'Junk' DNA not junk but key to complexity

There's a very nice video on Gene Regulation(free) from Science Magazine that discusses the pivotal roles that these non-coding regions of DNA play in our genome.

As John Mattick of The University of Queensland states at the end of the video:"We're just realizing that we've only got to first base and we have a long way to go, and most of the journey forward is going to be dissecting, analyzing and rebuilding an understanding of the massively parallel and extremely sophisticated RNA regulatory circuits, which really do underpin our complexity. And the irony, I think, is that what was dismissed as junk, because it wasn't understood, will turn out to hold the secret of human complexity, including our cognitive complexity. And that's where were going over the next 10 to 15 years.

More info on John Mattick's work is provided by a News in Science article (May 10,2004):The researchers scanned the human, rat and mouse genomes for matching regions of 200 or more DNA base pairs and found 481 regions that were completely unchanged. They then looked at earlier organisms.

"We then looked at the dog and bovine genomes and found that they were preserved there. Amazingly, most of them were preserved in the chicken genome, which has just been released, and about half are preserved in fish," Mattick said. "So that means some of these sequences have remain unchanged during evolution for over 400 million years."

Mattick said that these sequences remained unchanged while protein-coding genes changed slowly through evolution.

"So whatever [these conserved regions] are, and whatever they're doing, evolution is really saying that they're critical to our biology in ways that we don't yet understand."

Mattick said some of the sequences overlapped with protein-coding genes, while some were outside genes. But all were strongly associated with genes involved in controlling development. "They're almost certainly regulatory," he said.

The blog post on RNA interference provides some further details on the role RNA plays in transcriptional gene regulation. Additional info on RNA splicing in dendrites is provided in the blog post on dendritic spines. But there's a lot more going on here...

For one thing, function specific proteins can be 'stockpiled' in 'cytoplasmic granules', as well as sent to these granules for destruction. From The Scientist - A New View of Translational Control (Dec. 5, 2005): "Researchers are rapidly uncovering so-called granules in the cytoplasm that cluster function-specific proteins for RNA storage, silencing, reuse, destruction, and perhaps even splicing. Apparently related to the well characterized maternal mRNA granules that jumpstart embryogenesis, these neighborhood processing centers serve important functions in adult cells, including shaping synaptic plasticity and responding to stress.
"I think we'll learn that how cells control the destruction and translation of messenger RNAs through these structures will be a fundamental part of the control of genetic expression," says Roy Parker at the University of Arizona in Tucson. In the past two years, Parker has found cytoplasmic structures containing mRNA decapping and degradation enzymes. These compartments first appeared to serve as an mRNA junkyard: Transcripts with shortened poly(A) tails, or those otherwise no longer needed were relegated here for destruction. Parker dubbed them processing bodies, or P-bodies.
...
"It makes sense to have compartments for degradation. It's not just RNA randomly floating around with an enzyme happening to find it," says Keith Blackwell at Joslin Diabetes Center in Boston. But P-bodies may be more than just centralized paper shredders; they may store mRNA for later use. In September, when Parker and colleagues blocked translation in yeast cells by depriving them of glucose, the number of free-floating ribosome complexes known as polysomes decreased, and P-bodies grew in size as mRNAs went to them.5 But instead of being degraded, mRNAs accumulated. When glucose was restored, P-body size decreased and polysome number rose, suggesting that mRNAs were getting reused for translation. Reusing old mRNAs is likely more efficient and faster than making new ones, says John Rossi at the Beckman Research Institute of the City of Hope in Duarte, Calif.
...
In neurons, mRNA granules seem to influence synaptic plasticity (the variability in a synapse's signal strength), which appears fundamental to memory formation and learning. Kosik and colleagues found that granules store translationally silent mRNAs in dendrites. When the cell is depolarized, Kosik hypothesizes that the granules release their mRNAs to polysomes, resulting in localized protein changes. "They make sure that translation is directed to specific locations and not in the wrong place," he explains. The importance of such systems is hard to predict, Kosik says: "We could be talking about a branch of biology as extensive and intricate as the study of how proteins are directed to their destinations." Parker notes that neuronal and maternal granules have proteins in common and says he's looking to see if neuronal granules also possess P-body proteins.

More recently, James H. Eberwine of U.Penn reports in his web page: We have shown that multiple mRNAs are localized in neuronal dendrites and have provided a formal proof of local mRNA translation in dendrites. Further, we have recently shown that the intracellular sites of localization and translation of these mRNAs can be altered by synaptic stimulation highlighting for the first time that in vivo translation of a mRNA can occur at different rates in distinct regions of a single cell (translation is primarily exponential in dendrites and linear in the cell soma).

More info on the work of Eberwine and colleagues is described in an article in The Medical News: RNA-associated introns guide nerve-cell channel production:In nerve cells, some ion channels are located in the dendrite, which branch from the cell body of the neuron. Dendrites detect the electrical and chemical signals transmitted to the neuron by the axons of other neurons. Abnormalities in the dendrite electrical channel are involved in epilepsy, neurodegenerative diseases, and cognitive disorders, among others.

Introns are commonly looked on as sequences of "junk" DNA found in the middle of gene sequences, which after being made in RNA are simply excised in the nucleus before the messenger RNA is transported to the cytoplasm and translated into a protein. In 2005, the Penn group first found that dendrites have the capacity to splice messenger RNA, a process once believed to only take place in the nucleus of cells.

Now, in the current study, the group has found that an RNA encoding for a nerve-cell electrical channel, called the BK channel, contains an intron that is present outside the nucleus. This intron plays an important role in ensuring that functional BK channels are made in the appropriate place in the cell.

When this intron-containing RNA was knocked out, leaving the maturely spliced RNA in the cell, the electrical properties of the cell became abnormal. “We think the intron-containing mRNA is targeted to the dendrite where it is spliced into the channel protein and inserted locally into the region of the dendrite called the dendritic spine. The dendritic spine is where a majority of axons from other cells touch a particular neuron to facilitate neuronal communication” says Eberwine. “This is the first evidence that an intron-containing RNA outside of the nucleus serves a critical cellular function.”

“The intron acts like a guide or gatekeeper,” says Eberwine. “It keys the messenger RNA to the dendrite for local control of gene expression and final removal of the intron before the channel protein is made. Just because the intron is not in the final channel protein doesn't mean that it doesn't have an important purpose.”

The group surmises that the intron may control how many mRNAs are brought to the dendrite and translated into functional channel proteins. The correct number of channels is just as important for electrical impulses as having a properly formed channel.

The investigators believe that this is a general mechanism for the regulation of cytoplasmic RNAs in neurons. Given the central role of dendrites in various physiological functions they hope to relate this new knowledge to understanding the molecular underpinnings of memory and learning, as well as components of cognitive dysfunction resulting from neurological disease.

So it really seems that each dendrite is remarkably self-contained, with its own mitochondrial energy supply, the ability to synthesize proteins and the ability to wharehouse proteins. And that the dendritic machinery can be dynamically reconfigured by the neuron based on synaptic activity - the mitochondria and the mRNA localization and translation sites can move from quiescent dendrites to active ones on demand.

Junk DNA has other secrets that are being discovered, as well - for example, RNA-guided mechanisms underlying genome rearrangement. From a recent article in ScienceDaily (May 21, 2009): Laura Landweber and other members of her team are researching the origin and evolution of genes and genome rearrangement, with particular focus on Oxytricha because it undergoes massive genome reorganization during development.

In her lab, Landweber studies the evolutionary origin of novel genetic systems such as Oxytricha's. By combining molecular, evolutionary, theoretical and synthetic biology, Landweber and colleagues last year discovered an RNA (ribonucleic acid)-guided mechanism underlying its complex genome rearrangements.

"Last year, we found the instruction book for how to put this genome back together again -- the instruction set comes in the form of RNA that is passed briefly from parent to offspring and these maternal RNAs provide templates for the rearrangement process," Landweber said. "Now we've been studying the actual machinery involved in the process of cutting and splicing tremendous amounts of DNA. Transposons are very good at that."
...
They have concluded that the genes spur an almost acrobatic rearrangement of the entire genome that is necessary for the organism to grow.

It all happens very quickly. Genes called transposons in the single-celled pond-dwelling organism Oxytricha produce cell proteins known as transposases. During development, the transposons appear to first influence hundreds of thousands of DNA pieces to regroup. Then, when no longer needed, the organism cleverly erases the transposases from its genetic material, paring its genome to a slim 5 percent of its original load.

"The transposons actually perform a central role for the cell," said Laura Landweber, a professor of ecology and evolutionary biology at Princeton and an author of the study. "They stitch together the genes in working form." The work appeared in the May 15 edition of Science.

Listen to your junk man - he's singing!

Connecting the dots... "Let us begin anew"

As I've learned more about bio-systems, starting from water molecules and working up to synapses and networks of neurons, I've come to appreciate how incredibly powerful and compact the molecular computing substrate that life is built on top of is. Our most powerful supercomputers take days to calculate how one protein molecule folds, when the simplest bacteria can perform millions of these operations in parallel in seconds. What these simulations give us, however, is insight into exactly what special characteristics each of the proteins has in all of the various shapes it can assume. Building up from this low level understanding, hopefully we will be able to understand what the larger-scale purpose is for each of the various signaling chains and genetic transcriptions that are taking place, and perhaps we may one day be able to model these complex molecular interactions using state machines and logic that allows us to achieve a functionally equivalent set of operations without having to precisely simulate cells at the molecular level.

There are a number of new approaches to try to get to this level of understanding.
On the BrainScience podcast mentioned in the previous post, Seth Grant provided some nice descriptions of the difference and connections between the "trendy" terms "genetics", "genomics" and "proteomics":
Genetics is the study of gene function or the function of the biology as revealed by genes, and typically involves the study of cells or animals where there has been a mutation or an abnormality introduced into a gene and as a result of that, the function of the cell or animal is changed. And, of course, the readers will understand this, but a mutation in a gene effectively means a change in the DNA sequence that encodes that gene.
Genomics is a different thing. Genomics is the study of the organization of all of the DNA or the 'genome'. And, of course, the genome encodes roughly 20,000 genes in mammalian systems, and therefore, when one is studying the genomics of man or mouse, we're studying all of the genes. Typically in genetics you might only study one gene at a time in many cases. So that gives you a sense of the difference between the large scale features of genomics and the somewhat small scale features of genetics.
Proteomics is the study of the sets of proteins, or all of the proteins that perform biological functions or are found in cells or tissues. "Proteome" is to proteins what "genome" is to genes. Again, proteome is dealing with large sets of molecules. In our case, we were particularly interested in the 'proteome' (or all of the proteins) found in synapses. But you might be interested in all of the 'proteome' of red blood cells, in other words, all of the proteins that are found in a red blood cell.

There's a very good paper called "The Many Facets of Natural Computing" that looks at some of the interaction networks that are active in biological systems. The paper was written by

Lila Kari, Department of Computer Science, University of Western Ontario, London, ON, N6A 5B7, Canada, lila@csd.uwo.ca

Grzegorz Rozenberg, Leiden Inst. of Advanced Computer Science, Leiden University, Niels Bohrweg 1, 2333 CA Leiden, The Netherlands, Department of Computer Science, University of Colorado at Boulder, Boulder, CO 80309, USA, rozenber@liacs.nl

Their copyright notice:
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
...
[A]t the cell level, scientific research on organic components has focused strongly on four different interdependent interaction networks, based on four different “biochemical toolkits”: nucleic acids (DNA and RNA), proteins, lipids, carbohydrates, and their building blocks.

The genome consists of DNA sequences, some of which are genes that can be transcribed into messenger RNA (mRNA), and then translated into proteins according to the genetic code that maps 3-letter DNA segments into amino acids. A protein is a sequence over the 20-letter alphabet of amino acids. Each gene is associated with other DNA segments (promoters, enhancers, or silencers) that act as binding sites for proteins which activate or repress the gene’s transcription. Genes interact with each other indirectly, either through their gene products (mRNA, proteins) which can act as transcription factors to regulate gene transcription – either as activators or repressors –, or through small RNA species that directly regulate genes.

These gene-gene interactions, together with the genes’ interactions with other substances in the cell, form the most basic interaction network of an organism, the gene regulatory network. Gene regulatory networks perform information processing tasks within the cell, including the assembly and maintenance of the other networks. Research into modeling gene regulatory networks includes qualitative models such as random and probabilistic Boolean networks, asynchronous automata, and network motifs.(ref.)
...
Proteins and their interactions form another interaction network in a cell, that of biochemical networks, which perform all mechanical and metabolic tasks inside a cell. Proteins are folded-up strings of amino acids that take three-dimensional shapes, with possible characteristic interaction sites accessible to other molecules. If the binding of interaction sites is energetically favourable, two or more proteins may specifically bind to each other to form a dynamic protein complex by a process called complexation. A protein complex may act as a catalyst by bringing together other compounds and facilitating chemical reactions between them. Proteins may also chemically modify each other by attaching or removing modifying groups, such as phosphate groups, at specific sites. Each such modification may reveal new interaction surfaces.

There are tens of thousands of proteins in a cell. At any given moment, each of them has certain available binding sites (which means that they can bind to other proteins, DNA, or membranes), and each of them has modifying groups at specific sites either present or absent. Protein-protein interaction networks are large and complex, and finding a language to describe them is a difficult task. A significant progress in this direction was made by the introduction of Kohn-maps, a graphical notation that resulted in succinct pictures depicting molecular interactions. Other approaches include the textual biocalculus, or the recent use of existing process calculi (π-calculus), enriched with stochastic features, as the language to describe chemical interactions. (ref.)

Yet another biological interaction network, and the last that we discuss here, is that of transport networks mediated by lipid membranes. Some lipids can self-assemble into membranes and contribute to the separation and transport of substances, forming transport networks. A biological membrane is more than a container: it consists of a lipid bilayer in which proteins and other molecules, such as glycolipids, are embedded. The membrane structural components, as well as the embedded proteins or glycolipids, can travel along this lipid bilayer. Proteins can interact with free-floating molecules, and some of these interactions trigger signal transduction pathways, leading to gene transcription. Basic operations of membranes include fusion of two membranes into one, and fission of a membrane into two. Other operations involve transport, for example transporting an object to an interior compartment where it can be degraded. Formalisms that depict the transport networks are few, and include membrane systems described earlier, and brane calculi.

The gene regulatory networks, the protein-protein interaction networks, and the transport networks are all interlinked and interdependent. Genes code for proteins which, in turn, can regulate the transcription of other genes, membranes are separators but also embed active proteins in their surfaces. Currently there is no single formal general framework and notation able to describe all these networks and their interactions. Process calculus has been proposed for this purpose, but a generally accepted common language to describe these biological phenomena is still to be developed and universally accepted. It is indeed believed that one of the possible contributions of computer science to biology could be the development of a suitable language to accurately and succinctly describe, and reason about, biological concepts and phenomena.


One of the problems that happens in science is that, in order to understand things deeply, scientists typically need to specialize in one specific area of research. As Daphne Koller, a professor of computer science at Stanford University, relates in an interview about her being awarded the first-ever ACM-Infosyst Foundation Award in Computing Sciences(ref.):
The world is very complex: people interact with other people as well as with objects and places. If you want to describe what’s going on, you have to think about networks of things that interact with one another. We’ve found that by opening the lens a little wider and thinking not just about a single object but about everything to which it’s tied, you can reach much more informed conclusions.

[Interviewer]Which was an insight you brought to the field of artificial intelligence…
Well, I wasn’t the only one involved. There had been two almost opposing threads of work in artificial intelligence: there were the traditional AI folks, who grew up on the idea of logic as the most expressive language for representing the complexities of our world. On the other side were people who came in from the cognitive reasoning and machine learning side, who said, “Look, the world is noisy and messy, and we need to somehow deal with the fact that we don’t know things with certainty.” And they were both right, and they both had important points to make, and that’s why they kept arguing with each other.

How did probabilistic relational modeling help settle the dispute?
The synthesis of logic and probability allows you to learn this type of holistic representation [of complex systems] from real-world data. It gives you the ability to learn higher-level patterns that talk about the relationships between different individuals in a reusable way.

You’ve begun applying your techniques to the field of biology.
Originally, it was a method in search of a problem. I had this technology that integrated logic and probability, and we had done a lot of work on understanding the patterns that underlay complex data sets. Initially, we were looking for rich data sets to motivate our work. But I quickly became interested in the problem in and of itself.

What problem is that?
Biology is undergoing a transition from a purely experimental science — where one studies small pieces of the system in a very hypothesis-driven way — to a field where enormous amounts of data about an entire cellular system can be collected in a matter of weeks. So we’ve got millions of data points that are telling us very important insights, and we have no idea how to get at them.

What have you learned about interdisciplinary collaboration from your work with biologists?
The important thing is to set up a collaborative effort where each side respects
the skills, insights, and evaluation criteria of the other. For biologists
to care about what you build, you need to convince them that it actually produces good biology. You have to train yourself to understand what things they care about, and at the same time you can train them in the methods of your community.

So it’s not just learning a new scientific language, but training yourself to respect a different research process.
It’s a question of finding people who are capable of learning enough of the other side’s language to make the collaboration productive.

This sentiment is echoed in numerous papers I've come across, as well as in the poetic conclusion of "The Many Facets of Natural Computing":
In these times brimming with excitement, our task is nothing less than to discover a new, broader, notion of computation, and to understand the world around us in terms of information processing.

Let us step up to this challenge. Let us befriend our fellow the biologist, our fellow the chemist, our fellow the physicist, and let us together explore this new world. Let us, as computers in the future will, embrace uncertainty. Let us dare to ask afresh: “What is computation?”, “What is complexity?”, “What are the axioms that define life?”.

Let us relax our hardened ways of thinking and, with deference to our scientific forebears, let us begin anew.

More...
Bulletin of the EATCS (2007): Machines of systems biology.
Nature (Sept. 2002): Cellular abstractions: Cells as computation
Cambridge University Press (1999): Computing and Mobile Systems - the π-Calculus
Information Technology in Systems Biology (Kohn Maps)
Developmental Biology (2007):The regulatory genome and the computer
Science Signaling (2004):Molecular interaction map of the mammalian cell
cycle control and DNA repair systems."
The Calculus of Looping Sequences for Modeling Biological Membranes"
IEEE (2007): A Uniform Framework of Molecular Interaction for an Artificial Chemistry with Compartments

"Once more into the breach, dear friends, once more!"

The more I read about "Cognitive Computing", the more disenchanted I get with most of the work being done under this banner. There is an awful lot of hype going on here: everything from university researchers that claim how simple it is to create a silicon chip that accurately emulates millions of neurons and projects to create silicon prosthetics for some of the major centers in the brain to overly ambitious claims stating how close we are to getting computers to 'think' and thus to the resulting 'singularity'. Most 'cognitive computing' efforts seem to miss the point that there is more happening here than simple electrical signaling over a network. So coming across the following articles and podcast was like a breath of fresh spring air:

Complex Synapses Drove Brain Evolution:

ScienceDaily (June 9, 2008) — One of the great scientific challenges is to understand the design principles and origins of the human brain. New research has shed light on the evolutionary origins of the brain and how it evolved into the remarkably complex structure found in humans.

The research suggests that it is not size alone that gives more brain power, but that, during evolution, increasingly sophisticated molecular processing of nerve impulses allowed development of animals with more complex behaviours. The study shows that two waves of increased sophistication in the structure of nerve junctions could have been the force that allowed complex brains - including our own - to evolve. The big building blocks evolved before big brains.

Current thinking suggests that the protein components of nerve connections - called synapses - are similar in most animals from humble worms to humans and that it is increase in the number of synapses in larger animals that allows more sophisticated thought. "Our simple view that 'more nerves' is sufficient to explain 'more brain power' is simply not supported by our study," explained Professor Seth Grant, Head of the Genes to Cognition Programme at the Wellcome Trust Sanger Institute and leader of the project. "Although many studies have looked at the number of neurons, none has looked at the molecular composition of neuron connections. We found dramatic differences in the numbers of proteins in the neuron connections between different species".

"We studied around 600 proteins that are found in mammalian synapses and were surprised to find that only 50 percent of these are also found in invertebrate synapses, and about 25 percent are in single-cell animals, which obviously don't have a brain." Synapses are the junctions between nerves where electrical signals from one cell are transferred through a series of biochemical switches to the next. However, synapses are not simply soldered joints, but miniprocessors that give the nervous systems the property of learning and memory. Remarkably, the study shows that some of the proteins involved in synapse signalling and learning and memory are found in yeast, where they act to respond to signals from their environment, such as stress due to limited food or temperature change.

"The set of proteins found in single-cell animals represents the ancient or 'protosynapse' involved with simple behaviours," continues Professor Grant. "This set of proteins was embellished by addition of new proteins with the evolution of invertebrates and vertebrates and this has contributed to the more complex
behaviours of these animals.

"The number and complexity of proteins in the synapse first exploded when muticellular animals emerged, some billion years ago. A second wave occurred with the appearance of vertebrates, perhaps 500 million years ago."
...


There's an excellent podcast interview with Dr. Seth Grant at BrainScience - episode 51 that covers this work in more depth. Highly recommended!
Excerpt:
The ancestral proteins that are found in unicellular animals are the proteins that are found in more or less all of the different synapses in the brain of the mouse. The most recently evolved proteins - the vertebrate proteins - those are the ones that are most diverse in the brain regions of the mouse. So some of those proteins are very high, for example, in the frontal cortex, others might be high in the hippocampus, others might be high in the cerebellum; in other words, they're very variable like that.

So what that is telling us, then, and I'm just returning now to that ancient vertebrate synapse that arose before big brains, it tells us that when this 'big synapse' evolved, what the vertebrate brain then did as it grew bigger and evolved afterwards - it exploited the new proteins that had evolved into making new types of neurons in new types of regions of the brain.

In other words, we would like to put forward the view that the synapse evolution has allowed brain specialization - regionalization - to occur. And we know from many many studies that the regionalization of the brain - there's parts involved with learning, there's parts involved with fear, there's parts involved with some aspect of mood or so on, there's parts involved with motor function - that all appears to be built on the template of molecular evolution of the synapse. "

Journal References
Nature Neuroscience, 8 June 2008 Evolutionary expansion and anatomical specialization of synapse proteome complexity.
Emes RD, Pocklington AJ, Anderson CNG, Bayes A, Collins MO, Vickers CA, Croning MDR,
Malik BR, Choudhary JS, Armstrong JD and Grant SGN.

PubMed Abstract:Neurotransmitters drive combinatorial multistate postsynaptic density networks.
Coba MP, Pocklington AJ, Collins MO, Kopanitsa MV, Uren RT, Swamy S, Croning MD, Choudhary JS, Grant SG.

The mammalian postsynaptic density (PSD) comprises a complex collection of approximately 1100 proteins. Despite extensive knowledge of individual proteins, the overall organization of the PSD is poorly understood. Here, we define maps of molecular circuitry within the PSD based on phosphorylation of postsynaptic proteins. Activation of a single neurotransmitter receptor, the N-methyl-D-aspartate receptor (NMDAR), changed the phosphorylation status of 127 proteins.

Stimulation of ionotropic and metabotropic glutamate receptors and dopamine receptors activated overlapping networks with distinct combinatorial phosphorylation signatures. Using peptide array technology, we identified specific phosphorylation motifs and switching mechanisms responsible for the integration of neurotransmitter receptor pathways and their coordination of multiple substrates in these networks. These combinatorial networks confer high information-processing capacity and functional diversity on synapses, and their elucidation may provide new insights into disease mechanisms and new opportunities for drug discovery.

Infomax

From "Modeling the Mind: From Circuits to Systems: section 1.2 "Sensory Coding" by Suzanna Becker.
"Several classes of computational models have been influential in guiding current thinking about self-organization in sensory systems. These models share the general feature of modeling the brain as a communication channel and applying concepts from information theory. The underlying assumption of these models is that the goal of sensory coding is to map the high-dimensional sensory signal into another (usually lower-dimensional) code that is somehow optimal with respect to information content. Four information-theoretic coding principles will be considered here: 1) Linsker's Infomax principle, 2) Barlow's redundancy reduction principle, 3) Becker and Hinton's Imax principle, and 4) Risannen's minimum description length (MDL) principle. Each of these principles has been used to derive models of learning and has inspired further research into related models at multiple stages of information processing.
...
The Infomax principle has been highly influential in the study of neural coding, going well beyond Linsker's pioneering work in the linear case. One of the major developments in this field is Bell and Sejnowski's Infomax-based independent Component Analysis (ICA) algorithm, which applies to nonlinear mappings with equal numbers of inputs and outputs (Bell and Sejnowkski, 1995).
...
The principle of preserving information may be a good description of the very earliest stages of sensory coding, but it is unlikely that this one principle will capture all levels of processing in the brain. Clearly, one can trivially preserve all the information in the input simply by copying the input to the next level up. Thus, the idea only makes sense in the context of additional processing constraints. Implicit in Linsker's work was the constraint of dimension reduction. However, in the neocortex, there is no evidence of a progressive reduction in the number of neurons at successively higher levels of processing.

---

Transcript of presentation by Ralph Linsker (IBM TJ Watson Research center)
(video, slides)
Lisker's presentation slot starts at the 50 minute mark in the video (aprox. 40% through). (Lisker's an excellent speaker, but his slides leave much to be desired!)


Slide 1. The search for organizing principals of brain function.

"My working belief is that one needs multiple organizing principles at multiple levels of the brain ranging from synapse up to hierarchies of areas within the neocortex and different areas apart from the neocortex. And my working belief is that the number of such high level organizing principles one might need is more than one but less than ten. And I'm going to talk about a couple of aspects of what these potential organizing principles might be, my special interest being at the level between cell and cortical maps.
...

Slide 2. Self-organization
It's striking to me sometimes how long it takes certain ideas to be put together, to be combined from different disciplines.
Turing had a wonderful paper, not the one for which he's most famous, but it's a seminal paper on morphogenesis in biology. 1952.
Hebb's idea that you've heard about dates from 1949.
An early puzzle in neuroscience arose from the work of Hubel and Wiesel, experimental work starting in 1960 which showed that, in cats and then later in monkeys one finds a layer of cells in which each cell responds selectively, preferrentially, to a local edge at some particular orientation. And that as you move across cortex, you find a pattern of the different preferred orientations. [see image at the top of this post] The puzzle was, "how does this come about"? They even found in monkey that this is present at birth, so it does not develop as a result of exposure to structured visual stimuli in that case.

What I found, my introduction to this area of self-organization in neural systems, was that if you combine the Hebb rule with short connections (known to exist in retina) and simply some random electrical activity that's at least locally correlated - so if one retinal ganglian cell is activate or seeing a bright spot, it's likely that at neighbouring cell is going to be seeing a bright spot as well, a portion of the same patch - that those ingredients alone can lead to orientation selective cells and also to their patterning within a cortical layer. By the way, that locally correlated electical activity prenatally was not known to exist at the time but was found a few years later experimentally.

Slide 3: Self-organization in cortical models
What you see [in the color image at the top of this post] is a pattern that I generated that, at the time, troubled me because it looked more complex than Hubel and Wiesel's pattern of orientation domains. The color coding reflects the preferred orientation of the cell, as represented by one of 5 colors here. And at the time, Hubel and Wiesel's patterns only referred to a coarser grained resolution - are you closer to a horizontal preference or closer to a vertical preference, for example. And those patterns looked rath, like fingerprints, meandering stripes.

When I came up with this I was troubled at first, but it turned out later that experimental work published after this came out by Blasdo and Salama revealled that the complexity of this pattern is, in fact, what's found in cortex, including the singularities where most or all of the colors meet at a point, which are now commonly called 'pinwheels'. Now that's a static view of the end of the optimization process using my model. Here's a movie thanks to Sirotia et al that uses a more elaborate version of this same model for self organization but is based on the same principles.

(Movie)

Staring from random orientation preferences (or lack of preference), you evolve using simple Hebbian-type learning rules to get this kind of resulting pattern.

Slide 4: Some higher-level properties that can result from Hebbian Learning
Now, that's well and good. What are some higher level properties that can result from Hebbian learning? To put it another way, if the Hebb rule for synapses is a good algorithm, what is it good for? What computational tasks is it good for? I'll illustrate with a couple of analogies that are really much stronger than metaphor - there's a mathematical base for them that's solid. But just to make the point quickly:
If you apply a Hebb rule to the synapses impinging on a given output cell, the cell can be regarded as a committee in which each member has a voting strength that's initially random, but he gets more votes each time his preference agrees with the final output of the committee.

What this does is it induces a consensus forming by the committee on a subset of issues that frequently come before it for consideration. So this committee becomes, for example, an orientation selective analyzing committee. On most questions that come to it, most local input patterns that come to it, it will have no strong opinion. Where there's an oriented edge, it will have a strong opinion, perhaps positive or negative.

So Hebb's rule induces a committee consensus and I extended this to the issue of an entire layer of cells that can interact in a competitive and cooperative manner, through lateral connections, and proposed what I call the "infomax" principle, which says:

Create a layer of cells connecting inputs to outputs.

The cells can compute any of a wide class of functions, subject to certain biological constraints.

Let it develop in such a way that its outputs convey maximum Shannon information on average about its inputs, subject to those biological constraints and costs.

the costs can be of different types: it could be types of allowed processing - how strong are the processors as computers, limited wire length, energy costs, and so forth.

It's an optimal encoding principle, and again, for a brief metaphor, imagine an organization now of human beings where no person is told what their job is explicitly, and in fact, no one is told what the goal of the entire organization is. All they're told is "you're going to receive masses of data each day, and your job is to write a summary in one page that captures as much Shannon information as possible about that input.

What each person will do is within the limits of their ability , find regularity, find patterns within that data so that they can capture it more concisely. And that, in essence, is what Infomax does.

It's been used in various ways, extensively for models of neural learning and development.

It leads to qualitative and quantitative agreement with experiment, especially in the first few stages of early visual processing.

It's the basis of Bell and Sejnowski's Independent Component Analysis method, which can reconstruct N statistically independent sources, given at least N linear combinations of them.

...


More...
Journal of vision article: Cone selectivity derived from the responses of the retinal cone mosaic to natural scenes
Andrei Cimponeriu (Georgetown Institute for Cognitive and Computational Sciences,
Georgetown University Medical Center):
Modeling the Development of Ocular Dominance and Orientation Preference Maps in The Primary Visual Cortex with The Elastic Net

"That which I cannot build, I do not truly understand" -- Richard Feynman

In 2006, IBM Research hosted a series of lectures on Cognitive Computing, featuring presentations from some well-known researchers in neuroscience and cognitive computing. Videos of the lectures and the presentations that were given are available at http://www.almaden.ibm.com/institute/2006/agenda.shtml. A word of caution, however: as one person in the audience commented in a Q&A session after a panel presentation, a number of the presentations were more 'neuromythology' (i.e. bravado, marketing, speculation and wishful thinking) than neuroscience. I did learn a number of things from a few of the presentations, however, and will try to summarize the good stuff and ignore the rest in the next few posts.

The presentation by Henry Markram, EPFL/BlueBrain: The Emergence of Intelligence in the Neocortical Microcircuit (video) describes the Blue Brain project that Markram was director of at the time, which aimed to create a computer model of the neurons in a cortical column using a supercomputer to model each neuron and networking over 8000 of these supercomputer nodes together using MPI (Message Passing Interface - an industry standard messaging protocol for parallel computing). "Phase 1" of this work was completed in 2007.

Markram and his team's work was a technological Tour de Force, tackling some incredibly daunting challenges head on (ahem). For this post, I'd like to narrow the focus to some of the things I learned about spiking neuron models from Markram's presentation, and link some of these concepts to some of the things covered in previous posts. The images below are from Markram's presentation.

Re: Perkinjes and Granules and Schwanns, oh my...
Each neuron is unique, but when you look at a large number of them (as you need to do when you contemplate trying to model a 10,000 neuron cortical column!) you start to see similarities between the various neurons, enough so that you can classify them by shape:


Each of these classes of neurons can exhibit a wide variety of electrical behaviours:

Re: Ion Channels: gates in the cell wall and Receptors: getting the message across:

One of the factors that determines the electrical behaviour of a neuron is the combination of ion channels that it supports. You can determine which ion channels a particular neuron has 'implemented' by harvesting the neuron's cytoplasm, extracting the mRNA strands, performing reverse transcription and identifying all of the genes that code for ion channels.



Re: Will you remember me? I will remember you...
For all of the amazing fidelity and accuracy of the neuron models being used to create Blue Brain, there are a number of things that it doesn't tackle: e.g. the internal cellular biology of the neurons and the ability of a neuron to grow or modify its dendritic spines. At this stage ("Phase 1"), the Blue Brain project focused on creating a static snapshot in time of the neurons in the cortical column.

From the Blue Brain FAQ:

Q: How will you be able to replicate the complexity of neurons and neurotransmitter actions?

A: We have built 3D computer models of most of all the main types of neurons and can simulate their individual behaviors with great detail and very accurately. At this stage we can capture the complexity of the fast neurotransmitters very accurately as well with phenomenological models that we have built. A more difficult issue is the slow neurotransmitters and the neuromodulators as well as hormonal effects. These will take a while longer to model, but there is no major obstacle to this.

Q: What is the difference between cellular and molecular simulation?

A: The cellular level is a form of phenomenological model of the underlying molecular processes - a simplification - so it does capture many key processes, but molecular interactions are of course very complex and they keep neurons on a growth trajectory (real neurons are never biochemically stable), whereas in the simulations, neurons will tend to go back to a resting position when not activated. A very important reason for going to the molecular level is to link gene activity with electrical activity. Ultimately, that is what makes neurons become and work as neurons - an interaction between nature and nuture.


Two other questions that the FAQ doesn't address: What will "Phase 2" focus on and when will it get underway? A couple of news items provide a bit of a glimpse of what's next:

From IEEE Spectrum's TechTalk:
David Cremese, the manager of Deep Computing Programs at IBM Zurich, told me that the first phase of Markram's project is complete but that IBM intends very much to collaborate on future phases.

From TechnologyReport:
Technology Report has confirmed with IBM Switzerland that the Blue Brain project is waiting for Phase II funding from the Swiss Government. See the statement from Blue Brain project director Henry Markam ... as quoted by IBM Switzerland to Technology Report on January 19, 2009:

The funding:
There is a serious misconception that IBM somehow funded or donated to support the Blue Brain Project. The BBP project is funded primarily by the Swiss government and secondarily by grants and some donations from private individuals. The EPFL bought the BG, it was not donated to the EPFL. It was at a reduced cost because at that stage it was still a prototype and IBM was interested in exploring how different applications will perform on the machine - we were a kind of beta site.

The Collaboration:
The Blue Brain Project is a project that I conceived over the past 15 years. I chose the name because of the Blue Gene series which is a fantastic architecture for brain simulations. When we bought the BG, we also had to make sure that we have the computer engineering and computer science expertise to run the machine and optimize all the programs. So BG came to us with IBM’s full support as a technology partner. This component of the collaboration is invaluable to the Project and will continue and grow as long as we have a Blue Gene or other architectures from IBM. This
is by far the major component of the collaboration.

IBM Research at T.J. Watson, also contributed a postdoc that was sent to work with us at the EPFL and assigned a researcher at Watson to work on some computational neuroscience tasks. The research and term assigned to these postdocs is done, a success and published. Actually, the term expired almost a year ago, and the IBM postdoc, Sean Hill, actually transfered and is now an employee of the BBP and not IBM. The researcher at TJ Watson worked on a specific problem of collision detection between the axons and dendrites and this is done very well and already published. Although very important projects and contributions, this is a small part of the BBP which is carried out at the EPFL and involves, neuroscience, neuroinformatics,
vizualization, and a vast spectrum of computational neuroscience.

BBP needs BG’s to continue the project. The architecture is perfect for brain simulations. When we manage to get our funding to buy the next BG/P finalized, we will start Phase 2 and that will of course involve the basic (and most significant) technology collaboration, and most likely also many new collaborations on specific research targeted topics where we see that IBM can, and would like to, contribute. So this is an intermediate phase while we get ready for phase 2 - molecular level modeling.

BBP sees IBM as a key partner in the BBP and I do think that IBM also sees the value in the BBP. We are getting ready for Phase 2, but it has not started until we get the next BG series.

One further hint as to what Markram might be thinking about for Phase 2 is alluded to in an aside Markram made on "Microcircuit plasticity" 48 minutes into his presentation (slide 56):
"We patched 6 cells, and we see how they're connected so we can define the circuit [they make]. Now we take the pipettes out and we wait 12 hours, and we re-patch it. And what we found is that the circuit was different. Not only after 12 hours but actually after 4 hours.

And just to show you how much inertia there is in the current scientific paradigm, [Science magazine] said that this was not interesting. It will come out in PNAS in another 2 months.". (aside: some interesting comments on this work here, including a reference to the PNAS paper). Markram continued: So we do these recordings, and we puff glutamate now [into the circuit] - we actually activate the circuit. We can't still put intelligent stimulus, but we activate the circuit. And when you activate the circuit, here you can see that you have connections appearing and disappearing. This is potentially the substrate that Nobelist Gerry Edelman could use in all kinds of restructuring of the circuitry. Over a 4 hour period you can still see the circuitry is dynamically rewiring. For 50 years we've studied only how synapses are getting stronger and weaker, not how the circuit restructures itself.


This, to me, is one of the most important "forward looking" things Markram focused on during the talk, because it goes beyond the idea of modeling the brain as a static 3D electrical network made up of ion channels and opens the door to the idea that the protein synthesis, dendritic spine growth and neuron rewiring that have been observed to happen with real neurons are also important factors in how the brain works. I hope this is a hint of things to come in Phase 2!

A couple of thoughts occurred to me was as I was going through the presentation. One was that what neurons are really designed to do is to precisely send chemical signals to a specific set of other cells. Instead of releasing chemical messenger molecules out into the body where any cell can pick them up, neurons extend long appendages that deliver the chemical messages right to the front door of the cells that it wants to receive the messages. How does it know which cell(s) to grow the appendages towards? One of the ways the body uses to govern how cells grow during development ('morphobiology') is through the use of chemical gradients that trigger genetic transcription factors at certain points along the gradient. I wonder what chemical (or 'electro-chemical?') gradients exist in the space between neurons that could guide this growth?

The second thought occurred to me after seeing the variety of different types of action potentials that the neurons can generate: perhaps the 'neural code' that these bursts of spikes transmit is not simply used to pass on information that has been received by the senses and processed by other regions in the brain - perhaps it is a set of instructions to the cells that this information is being passed on to that help these cells retain and process this information by generating the appropriate set of proteins at exactly the right time - an electrical stimulus to trigger the necessary chemical chain reactions within the cytoplasm of the cell. This would link up the work done by Dr. Fields et. al at the cellular and molecular level with the with work being done at the connectionist / action potential modeling level... Time to fire up Google and see what work has been done in this area!

Synchronicity - spatio-temporal spiking neuron models

The previous post began with a slogan pertaining to Hebbian learning that was coined by Donald Hebb: "Cells that fire together, wire together". A number of papers have been appearing in recent years that extend this idea further - that pulses that coincide are actually one of the most important ways that the brain transmits information. This concept appears to be a natural consequence of Hebbian learning: the brain adapts its network of synaptic connections by pruning those connections where the incoming signals are not correlated with other signals coming into the neuron and reinforces those where this type of coincidence does occur. It is doing this for a reason - to establish the 'right' set of connections and synaptic weights in order to associate one input or one set of inputs with another. This type of correlation between events has been proposed as being what knowledge itself is made of and as the basis for some of the key aspects of cognition and symbolic thought (ref.).

Not everyone agrees with this idea that the relative timing of spikes is what is used to carry information, however. There are many researchers that focus on trying to understand the 'neural code' that is used by the brain to send information from one neuron to another. Claude Shannon's Information Theory is often used as a mathematical framework to try to determine how many 'bits of data' are sent from one neuron to another and the efficiency of the information transferred from one neuron to the next. Researchers who favour this approach generally model the transmission of spikes from one neuron to the next using "rate coding" (a.k.a. "frequency coding") models. These models are based on the idea that neurons will only reach the threshold needed to generate a spike and send it to the next neuron when the number of spikes it receives exceeds in a given time period exceeds some value.

Although information theory has proven to be a useful way to gain insight into the lower levels of vision processing (e.g. Ralph Linsker's Infomax principle), it doesn't seem to me that constraining the theory of how neurons interact so that it can be described using the math of Information Theory is helpful - the concept of 'how much information' is transferred from one neuron to the next is determined primarily by which neuron the spike is sent to; information theory is not the right tool to approach this with. Using the mathematical framework of information theory narrows the focus down to things like the frequency of the spikes that are sent from one neuron to the next, and researchers end up making the wrong type of simplifications in order to achieve this - things like combining both the efficiency of the synapse and the 'frequency' of the incoming spikes into a single 'synaptic weight', and ignoring the idea that the network of neurons is constructed in such a way as to enable coincidence detection. A different way of measuring information is needed - one that takes into account the 100's, 1000s or 10,000s of connections each neuron can send that spike to, whether the spike is used to inhibit or excite the downstream neuron, the topology of the connections between groups of neurons, the fact that neurons can rewire themselves dynamically, the role that neurotransmitters play, etc.

The rate coding model aligns well with the popular Computer Science approaches to implementing neural networks in software - time does not play a significant role in the way these models operate. Training data is used to ensure that the weights converge to whatever is required in order to map the input data to the expected output data in the training set. What these software models completely ignore is the idea that the input signals in the brain are sent as neuronal spikes, and that the correlation between spikes is what causes the synapses to either grow stronger or weaker. Rate coding attempts to bridge this gap by noting that spikes from a neuron often are generated in a repetitive series - a spike train - and that the frequency of these spikes will tend to drive the receiving neuron beyond its signaling threshold. The relative density of spikes over a given time interval, the thinking goes, is what matters; the timing of the individual spikes is not important. Rate coding also glosses over the fact that neurons do not have spikes arriving at a constant rate - they arrive sporadically and in bursts (ref.)

The rank coding model (a.k.a. Delay coding), on the other hand, is based on the idea that sensory neurons (e.g. in the retina and inner ear) will respond to more energetic input signals by generating a spike earlier, that this generated spike will arrive at the downstream neuron earlier than other spikes and will thus have a higher impact or 'ranking' relative to later incoming spikes as a result. The rank coding model notes that there are some specific examples where the only way the brain can respond quickly enough to an incoming stimulus (e.g. a noise or an image) is if a single neuron were to respond to the initial spike that was sent from the neuron in the eye or ear. Software models of rank-encoding methods and "liquid state machines" are starting to appear (e.g. SpikeNET Technology ) which offer capabilities that outperform standard software neural networks for certain appliations. These are the leading edge of a "third generation" of software models of neural networks, which look very promising.

From Networks of Spiking Neurons: A New Generation of Neural Network Models by Thomas Natschläger (December 1998):

The First Generation of Models
If one wants to understand how the nervous system computes a function one has to think about how information about the environment or internal states is represented and transmitted. The fact that the shape of an action potential is always the same one can exclude the possibility that the voltage trajectory of an action potential carries relevant information. Thus a central question in the field of neuroscience is how neurons encode information in the sequence of action potential they emit. In this article we characterize neural network models by the assumptions about the encoding scheme.

1943 McCulloch and Pitts proposed the first neuron model: the threshold gate. The characteristics of their model was that they treated a neuron as a binary device. That is they distinguished only between the occurrence and absence of a spike. The threshold gate is used as building block for various network types including multilayer perceptrons, Hopfield networks and the Boltzman machine. It turned out that the threshold gate is a computational powerful device. That is one can compute complex functions with rather small networks made up of threshold gates. From the theoretical point of view the threshold gate is a very interesting model but it is unlikely that real biological systems use such a binary encoding scheme. A prerequisite for such a binary coding scheme is a kind of global clocking mechanism but it is very unlikely that such a mechanism exists in biological systems.

The Second Generation
Another possibility is that the number of spikes per second (called the firing rate) encodes relevant information. This idea lead to a model neuron known as sigmoidal gate. The output of a sigmoidal gate is a number which is thought to represent the firing rate of the neuron. There exists a huge amount of literature which discusses in detail all aspects of this kind of neural network models. We just want to note that networks of sigmoidal gates can in principle compute any analog function and that along with this type of models the question of learning in neural networks was intensively investigated for the first time.

...
The Third Generation: Networks of Spiking Neurons (SNN)
...[Results] from experimental neurobiology gave rise to a new class of neural network models where one also incorporates the timing of individual spikes. Thus time plays a central role in SNNs whereas in most other neural network models there is even no notion of time.


The standard Computer Science neural networks are based on the second generation of models, and have been found to be useful in applications including text recognition, speech recognition and stock market prediction. Software implementations of the third generation of models are starting to appear. They are based on more accurate computer representations of biological neural networks and will hopefully open up new types of software applications in areas such as machine vision and robotics.

References

Rate Codes and Shannon's Information Theory
Information theory and neural coding - Alexander Borst and Frédéric E. Theunissen
Energy-efficient interspike interval codes
Neural coding and decoding: communication channels and quantization
Introduction: statistical and machine learning based approaches to neurobiology - Shin Ishii
Nara Institute of Science and Technology
Information Theory and Systems Neuroscience
Entropy as an Index of the Informational State of Neurons

Rank Coding and Temporal Coding
SpikeNET - Scientific papers by Simon Thorpe & colleagues
Publications related to SpikeNET
The Neural Basis of Temporal Processing
Temporal Coding and Analysis of Spike Sequences
Synfire Chains and Cortical Songs: Temporal Modules of Cortical Activity

Spike-based Neural Network Models
Networks of Spiking Neurons: A New Generation of Neural Network Models - Thomas Natschläger
Computing with spikes - Wolfgang Maass

Neurotrophins

In 1949, Canadian psychologist Donald Hebb proposed that "When an axon of cell A is near enough to excite cell b or repeatedly and consistently takes part in firing it, some growth process or metabolic changes take place in one or both cells such that A's efficiency, as one of the cells firing B, is increased". (ref.) This idea is captured in the slogan 'Cells that fire together wire together'. A special set of molecules called neurotrophins play an important role in this. From Joseph LeDoux's book The Synaptic Self: When an action potential occurs in a postsynaptic cell, neurotrophins are released from the cell and diffuse backward across the synapse, where they are taken up by presynaptic terminals. Under the influence of neurotrophins, the terminals begin to branch and sprout new synaptic connections. Since only those presynaptic cells that were just active (that just released transmitter) take up the molecules, only they sprout new connections. activity thus induces growth, and the growth that occurs is restricted to the active terminals.
In addition to this role in the active construction of ciruits, neurotrophins are also involved in synapse selection. The natural fate of may cells during development is an early exit. So-called programmed cell death is one of the regressive events that help shape the final pattern of connectivity. Cell death is prevented if a presynaptic terminal receives a lifes-sustaining shot of neurotrophins from it postsynaptic partner. The survival rate of neurons is in this way regulated by the limited availability of neurotrophins. Only those cells that compete successfully for neurotrophins (those that are active) survive. In the presence of neurotrophins, the surviving terminals (those that were active) alsso begin to sprout new connections. Selection can be a step along the path toward activity-instructed growth -- in other words, selection and instruction are partners in synaptic development.
The image at the top of this post shows undifferentiated cells extending neuronal processes after exposure to neurotrophin delivered from hydrogel coated neural prosthetic devices. This idea of using neurotrophins to efficiently 'wire up' neurons with prosthetic devices is quite intriguing. Let's take a closer look at what neurotrophins are and how they work...

From the Society for Neuroscience article Neurotrophic Factors :
Recent research shows that neurotrophic factors are:

Present in early development of the nervous system and are responsible for the initial growth and development of neurons in the peripheral and central nervous systems.

Released by target tissue of a growing neuron and can determine whether a neuron reaches its target during development; neurons which do not reach the target die.

Capable of making damaged neurons regrow their processes in a test tube and in animal models and, thus, represent exciting possibilities for reversing devastating disorders, including Alzheimer's disease, Parkinson's disease and amyotrophic lateral sclerosis (ALS), also known as Lou Gehrig's disease.

...
Neurotrophic factors, produced by several body tissues including muscle, act by attaching to receptors on the tips, or nerve terminals, and on the cell body -- which contains the nucleus -- of neurons. The signal can then be carried through the axon, the neuron's elongated fiberlike extension, which can be as long as a yard, to the cell body where it tells the cell what to do.
Thus far, scientists have identified several neurotrophic factor receptors -- which also may be potential targets for therapy. A receptor called trk is required for the action of nerve growth factor (NGF), the first neurotrophic factor, which was discovered 40 years ago. NGF affects primarily neurons using the neurotransmitter acetylcholine in the basal forebrain, sensory neurons and sympathetic neurons that regulate organs such as the heart and lungs. Relatives of trk are receptors for other neurotrophic factors -- trkB seems to be a receptor for brain-derived neurotrophic factor; and trkC for neurotrophin-3.
...
In the brain, a neurotrophic factor is released by a neuron or a support cell, such as an astrocyte, and binds to a receptor on a nearby neuron. This binding results in the production of a signal which is transported to the nucleus of the receiving neuron where it results in the increased production of proteins associated with neuronal survival and function.


From Developmental Biology online" by Scott F. Gilbert:
The neurotrophin family consists of four members: nerve growth factor (NGF), brain derived neurotrophic factor (BDNF), neurotrophin 3 (NT-3), and neurotrophin 4 (NT-4). Each provides some survival activity on nervous tissue. As mentioned in the text, the final number of neurons innervating a particular organ is attained by thinning the population of neurons through programmed neuronal death. Here, neurotrophic factors secreted by cells in the target field protect the neurons from apoptosis (Korsching, 1993; Lewin and Barde, 1996). Thus, the final number of neurons innervating a target reflects the availability of neurotrophins. The ability of particular neural subsets to respond only to particular neurotrophins can explain the losses of certain peripheral sympathetic neurons in NGF-knockout mice, the deficiency of sensory neurons in BDNF knockout mice, the lack of proprioceptive neurons in NT-3 knockout mice, and the deficiency of particular sensory neurons in NT-4 knockout mice. Neurotrophins also play roles in regulating neuronal plasticity and in regulating the number of neural progenitor cells.

Neurotrophin receptors

There are two classes of neurotrophin cell-surface receptors. The p75 receptor (also known as the low-affinity neurotrophin receptor, LANR) is common to all members of the neurotrophin family. The high affinity receptors (having binding constants on the order of 10-11) include receptor tyrosine kinase proteins TrkA, TrkB, and TrkC. These receptors have different specificities for different members of the neurotrophin family TrkA is the receptor for NGF, trkB is the receptor for BDNF and NT-4, and trkC is the receptor for NT-3. However, NT-3 can also bind to trkA and trkB, but with lower affinity than to trkC, and with lower affinity than the primary ligands for these receptors. Similarly, NT-4 also binds to trkA but with lower affinity.

In addition to these "classical" receptors, the issue is complicated by the existence of isoforms of trkB and trkC, which lack the cytoplasmic tyrosine kinase catalytic region (Barbacid, 1995). These receptors are found throughout the developing body as well, and it is not known if these noncatalytic forms of the receptors act as agonists or inhibitors.

All four neurotrophins also bind to the low affinity nerve growth factor receptor, p75. The p75 receptor belongs to the tumor necrosis factor receptor family and was the first identified neurotrophin receptor (Johnson et al; 1986). This receptor will bind the neurotrophins, but it has no cytoplasmic tyrosine kinase domain (Chao and Hempstead, 1995; Greene and Kaplan, 1995; Segal and Greenberg, 1996). The roles of this receptor are controversial, as it may also be involved in either promoting or downregulating the response to the neurotrophin. P75 may function to increase the affinity of the trk receptors for their respective neurotrophins, or it may bind the neurotrophins and prevent them from binding to the high affinity receptors. Although it does not have a catalytic intracellular tyrosine kinase domain, it is capable of mediating the neurotrophin signals. The ligand binding of p75 increases the high-affinity TrkA binding sites, enhances TrkA autophosphorylation and selectivity for neurotrophin ligands (Kaplan and Miller, 1997). P75 also increases intracellular ceramide levels and further activates NFk B transcription factor (Carter et al., 1996) and JNK kinase (Casaccia-Bonnefil et al., 1996) independently of tyrosine kinase activity. Conversely, TrkA activation can inhibit p75-mediated signaling, but the mechanism of this inhibition is unclear (Kaplan and Miller, 1997).

The TrkA neurotrophin receptor has been linked to human diseases. The TrkA gene was originally described as an oncogene in colon cancer (Martin-Zanca et al., 1986) and its translocations are common in papillary thyroid carcinoma (Bongarzone et al., 1989). Recently, a mutation in the TrkA gene was found to cause congenital insensitivity to pain with anhidrosis (CIPA) syndrome (Indo et al., 1996) that closely resembles the phenotype of the TrkA -deficient mice. No disease associations have been described either for the TrkB gene, or the genes for p75NTR or any of the neurotrophins.

From the abstract for Neurotrophin secretion from hippocampal neurons evoked by long-term-potentiation-inducing electrical stimulation patterns by Annette Gärtner and Volker Staiger: The neurotrophin (NT) brain-derived neurotrophic factor (BDNF) plays an essential role in the formation of long-term potentiation (LTP). Their study found that instantaneous secretion of BDNF is evoked by the same type of action potentials that induce LTP, whereas stimuli that induce Long Term Depression of a neuron do not induce secretion of BDNF.

Recently, further studies have provided additional details on the workings of BDNF. From Backpropagating Action Potentials Trigger Dendritic Release of BDNF during Spontaneous Network Activity by Nicola Kuczewski et. al: We found that spontaneous backpropagating action potentials, but not synaptic activity alone, led to a Ca²⁺-dependent dendritic release of BDNF-GFP. Moreover, we provide evidence that endogenous BDNF released from a single neuron can phosphorylate CREB (cAMP response element-binding protein) in neighboring neurons, an important step of immediate early gene activation. Therefore, together, our results support the hypothesis that BDNF might act as a target-derived messenger of activity-dependent synaptic plasticity and development.

An article in Nature (The Yin and Yang of neurotrophins" by Bai Lu, Petti T. Pang & Newton H. Woo") provides insight into the role played by neurotrophins and how BDNF is synthesized in neurons. From this article, it appears that in some cases, instead of BDNF being released by the dendrites and promoting axonal branching, it instead can be released by the axon and promotes the growth of additional dendritic spines.

The following picture is a schematic showing the synthesis and sorting of brain-derived neurotrophic factor (BDNF) in a typical neuron.

First synthesized in the endoplasmic reticulum (ER) (1), proBDNF (precursor of BDNF) binds to intracellular sortilin in the Golgi to facilitate proper folding of the mature domain (2). A motif in the mature domain of BDNF binds to carboxypeptidase E (CPE), an interaction that sorts BDNF into large dense core vesicles, which are a component of the regulated secretory pathway. In the absence of this motif, BDNF is sorted into the constitutive pathway. After the binary decision of sorting, BDNF is transported to the appropriate site of release, either in dendrites or in axons. Because, in some cases, the pro-domain is not cleaved intracellularly by furin or protein convertases (such as protein convertase 1, PC1) (3), proBDNF can be released by neurons. Extracellular proteases, such as metalloproteinases and plasmin, can subsequently cleave the pro-region to yield mature BDNF (mBDNF) (4). MMP, matrix metalloproteinase.

The last picture in this post (also from (The Yin and Yang of neurotrophins" by Bai Lu, Petti T. Pang & Newton H. Woo") shows the role BDNF plays in promoting the growth of additional dendritic spines and how its absence results in the retraction of dendritic spines:

a) Molecular cascade of brain-derived neurotrophic factor (BDNF) processing in late-phase long-term potentiation (L-LTP). In response to theta-burst stimulation (TBS), tissue plasminogen activator (tPA) is secreted into the synaptic cleft and cleaves the extracellular protease plasminogen to yield plasmin (1). Plasmin then cleaves proBDNF (the precursor of BDNF, which is released in an activity-dependent manner), yielding mature BDNF (mBDNF) (2). mBDNF binds to TrkB and triggers a series of downstream signalling pathways to induce LTP (3). During the maintenance stage of LTP, mBDNF might be generated by intracellular cleavage after postsynaptic transcription and translation (4). By contrast, proBDNF secreted extracellularly remains uncleaved after low-frequency stimulation (LFS). Uncleaved proBDNF binds to the p75 neurotrophin receptor (p75NTR) (5) to facilitate the induction of long-term depression (LTD), possibly through the regulation of NMDA (N-methyl-D-aspartate) receptor NR2B subunit expression. b) Morphological alterations in synapses induced by pro- and mature BDNF. Left, BDNF–Trk signalling might be an active mechanism that converts activity-induced molecular signals into structural plasticity, contributing to synapse formation. Right, proBDNF–p75NTR signalling might be important in translating activity-dependent signals into negative modulation of structural plasticity, contributing to synapse retraction.

Some good progress is being made on using neurotrophins to encourage neurons to integrate with medical implants. From EurekAlert (with thanks to the Biosingularity blog): Plastic coatings could someday help neural implants treat conditions as diverse as Parkinson’s disease and macular degeneration.

The coatings encourage neurons in the body to grow and connect with the electrodes that provide treatment.Jessica O. Winter, assistant professor of chemical and biomolecular engineering at Ohio State University described the research Thursday, August 21 at the American Chemical Society meeting in Philadelphia. She is also an assistant professor of biomedical engineering.

Worldwide, researchers are developing medical implants that stimulate neurons to treat conditions caused by neural damage. Most research focuses on preventing the body from rejecting the implant, but the Ohio State researchers are focusing instead on how to boost the implants’ effectiveness.

“We’re trying to get the nerve tissue to integrate with a device — to grow into it to form a better connection,” Winter said.

She and her colleagues are infusing water-soluble polymers with neurotrophins, proteins that help neurons grow and survive.

They are combining different polymers, some shaped like tiny spheres and fibers, to create composite coatings that release neurotrophins in a steady dose over time. The coatings also give nerves a scaffold to cling to as they grow around an implant.

The researchers coated two kinds of electrodes — one, a flat electrode used in retinal implants, and the other a cylindrical electrode array used in deep brain stimulation. The first is being used in experimental treatments for macular degeneration, while the second holds promise for suppressing tremors in people who have Parkinson’s disease. So far, however, it appears that the neurite growth achieved in this manner is short-lived. More info on this is available here.