Dendritic plasticity and ‘input feature storage’

Any neuroscience textbook will tell you that learning and memory in the brain happens through synaptic plasticity, where the strength of the connections between neurons is modified. However, it’s now clear that there are also non-synaptic forms of activity-dependent plasticity, where neurons change their intrinsic properties without altering the synapses at all. This can happen either on a full-cell (global) or compartmental (local) scale within the neuron.

One use for global regulation of a cell’s excitability might be to maintain a particular average firing rate following changes in the input patterns. It’s been shown that some input-deprived neurons change their properties to become easier to spike, so that when they do eventually receive input they fire more action potentials than they would have before (see ‘homeostatic plasticity’ mention at Scholarpedia entry for Intrinsic Plasticity).

On the other hand, although local dendritic changes within neurons had been shown possible (e.g. ref 2), no one had really pushed the idea to demonstrate a real computational function. In March just gone Jeff Magee’s lab tried to tackle this in a Nature paper (main ref).

In an earlier paper (ref 3) they used two-photon glutamate uncaging to excite multiple synapses on an oblique branch of a rat CA1 pyramidal cell. If enough inputs were activated (~20) in a short enough time period (~6ms), fast Na+ based dendritic spikes were initiated. At the soma this looks like non-linear summation (the EPSP from multiple inputs is bigger than the sum of the individual responses). The effect isn’t that big (see figure G below), but it is something. Also, the local voltage change is likely to be much greater due to the high input impedance out in the thin dendrites. This could trigger things like synaptic plasticity, or spikes in other dendritic branches.

Unfortunately, dendritic spikes on their own are no longer news enough to get you into Nature. What Losonczy and co. found was that the cell could regulate this spiking on a branch-by-branch basis. So I could have a dendritic branch that doesn’t ever spike, but by applying carbachol or pairing the dendritic spikes with somatic action potentials, I could turn my ‘weak’ branch into a ’strong’ one (see figure below). They call this phenomenon branch-strength potentiation (BSP).

They also showed that BSP does not affect the individual synaptic responses, you get still get spikes even if you stimulate naive spines on the same branch, and neighbouring branches are not affected. This shows that it is something specific to the local dendritic membrane. Further experiments with knockout mice and pharmacology implied that BSP is mediated, at least in part, by a downregulation of A-type K+ channels (which would indeed render a branch more excitable).

Fair enough. To my theorist’s eye the data seem convincing – I believe in BSP (unlike ESP). My problem, however, lies in their conclusions. They claim that this is a plausible form of input feature storage for these (and maybe all other) neurons. In the supplemental info there is a schematic illustration of a scenario where this might happen in a real cell:

It’s worth looking through this figure. Parts a-f are supposed to form a little story. Part a shows a hypothetical CA1 pyramidal neuron with some weak daughter (blue) and strong parent (red) dendritic branches. The soma is the big white circle and there’s no axon shown. Parts b and c suppose that multiple inputs to the same weak dendrite (here suggested to be an array of CA3 place cells) might fire simultaneously, but because of the weak effect on somatic membrane potential would result in only poorly timed action potential output. However, if BSP is induced, the weak branch turns into a strong one (part d). The next time these same inputs arrive they cause a dendritic spike, which in turns triggers a reliable somatic action potential (part e). Hence all such CA1 pyramidal cells getting this type of input could ‘learn’ to robustly respond to certain stimulus features (part f).

So is this scenario likely to occur in a real animal? This question really strikes at the heart of the matter. Is BSP actually used by the brain or is it just an epiphenomenon? One issue is whether these dendritic spikes in this region actually occur in vivo. Losonczy et al estimate that to initiate a d-spike you need to activate about 20 individual synapses on a single branch, from a typical pool of maybe 200. That sounds feasible, but at the moment no-one really knows for sure. Also, d-spike initiation might be more difficult if there is a high level of background synaptic activity, which would make the membrane leakier, thus shunting any further synaptic input.

The next potential problem is the size of the effect. In part G of the first figure in this comment (above), we see that the non-linearity is rather weak. Of course, these measurements were made at the soma, so the effect may be quite marked at the dendrite itself but just gets attenuated as the signal travels to the soma. This might mean a huge voltage change in the dendrite, which could induce classic synaptic activity by releasing the magnesium block in synaptic NMDA receptors. They don’t report any evidence of this, but it would have been nice if it was at least explored a little, maybe through dendritic recordings or computational modelling. Regardless of what happens at the dendrite, since the amplitude of the effect at the soma is small it is hard to see how it could robustly control axonal spike timing (remember, the above figure is just a schematic).

A third complication is how to reconcile all of this with current models of pattern recognition. According to classic neural network theory, the ‘memory’ is stored solely in the synaptic weights. Although the individual synapses continue contribute similarly under this new scheme, their non-linear interaction is something that has not been explored much in the theoretical literature (although see work from Bartlett Mel’s lab, refs 4 & 5). One specific problem I can imagine comes from the fact that the dendritic spike doesn’t care which particular synapses initiated it (say any 20 of 200). Even if my dendrite had learned to spike because of one particular set of synchronous inputs, it would also interfere with any other ‘patterns’ stored in all the remaining synaptic weights on the same branch. It’s not clear how to handle this theoretically.

Overall, I think it’s great to see these kinds of ideas put forward by real experimentalists, even if they turn out not to be exactly correct. The huge flaw of most existing network models of learning and memory is that they ignore a lot of what we know about how real neurons operate. These inconveniences include the spatial extent of the dendritic tree, non-linear synaptic integration, and intrinsic plasticity. The days of the simple integrate-and-fire model of the neuron are long gone. A nice paper from Panayiota Poirazi’s lab (ref 6) summed it up when they said that “something smaller than the cell lies at the heart of neural computation”.

Main ref:
Attila Losonczy, Judit K. Makara, Jeffrey C. Magee (2008). Compartmentalized dendritic plasticity and input feature storage in neurons Nature, 452 (7186), 436-441 DOI: 10.1038/nature06725

Other refs:

1. Intrinsic Plasticity.
   Robert Cudmore, Niraj S Desai.
   Scholarpedia (2008) 3(2):1363.
   http://www.scholarpedia.org/article/Intrinsic_plasticity
2. LTP is accompanied by an enhanced local excitability of pyramidal neuron dendrites.
   Andreas Frick, Jeffrey Magee, Daniel Johnston.
   Nat Neurosci (2004) 7 (2), 126-35
   PMID: 14730307
3. Integrative properties of radial oblique dendrites in hippocampal CA1 pyramidal neurons.
   Attila Losonczy, Jeffrey C Magee.
   Neuron (2006)  50 (2), 291-307
   PMID: 16630839
4. Pyramidal neuron as two-layer neural network.
   Panayiota Poirazi, Terrence Brannon, Bartlett W Mel.
   Neuron (2003) 37 (6), 989-99
   PMID: 12670427
5. Computational subunits in thin dendrites of pyramidal cells.
   Alon Polsky, Bartlett W Mel, Jackie Schiller.
   Nat Neurosci (2004) 7 (6), 621-7
   PMID: 15156147
6. Inside the brain of a neuron.
   Kyriaki Sidiropoulou, Eleftheria Kyriaki Pissadaki, Panayiota Poirazi.
   EMBO Rep (2006) 7 (9), 886-92
   PMID: 16953202

Academic hype in neuroscience

Neuroscientists love to talk about the brain. Like all scientists, they feel the need to emphasise the link between their particular domain of expertise and the bigger picture. A neuroproteomics study might draw grand allusions to cognition, or a mathematics paper might speculate on the basis of drug-induced hallucinations.

The function of this hype is often to simply boost interest in the research: maybe to convince a hesitant editor of its importance, or to catch the eye of the academic trawling through the weekly table-of-contents email. For example, a recent neuroimaging study from Dick Passingham’s group carried the provocative title “Reading intentions in the human brain”. I do admit that these dramatics are at least understandable, if not usually justified. Researchers are often under pressure to publish work that will have benefits for humanity at large. Making these tenuous connections where possible might decide whether the next grant application is successful or not.

Beyond this, I sometimes feel that there’s still a more subtle motive at work. All the theatrics might be to purposely skew the reader’s opinion of a paper. Of course everyone wants to read interesting stuff. But what if I have nothing interesting to say? I could stay quiet. But maybe it would be better for me and my image to just say something, anything, then bathe it in a few buzzwords and round it off with some juicy speculation.

Let’s assume that I am presenting some valid, useful science which should stand up on its own. Why dress it up? I think that people often just want to make it publicly known that they too think about the big issues. It’s stealth advertising. It’s saying, I’m not just a fan of surfing; I’m part of the wave.

I was reminded of this last week when reading a couple of 1999 papers by Amit Manwani and Christof Koch (see refs belows). They attempted, with some success, to apply techniques from information theory to biophysical signal transmission within single neurons. The basic idea was to calculate how much information is lost as the electrical signal passes from the synapse to the soma along a dendrite, due to the interference of neural noise. I thought it was nice science, but a little buzzword heavy. This is a common irk I have with these information-theoretic-neural-coding papers. Here’s an excerpt from the second paper:

We are interested in deconstructing neuronal information transfer into its constituent biophysical components and assessing the role of each stage in this context rather than arriving at an accurate estimate of neuronal capacity.

It’s not that there’s anything particularly incorrect about the sentence. It’s just that it’s carefully constructed to include all the eye-catching jargon that goes into your typical Neural Computation article. Maybe I’m being a bit harsh on Manwani and co. – I really think it’s great to see some biophysics in the mix. Much better than the usual info-theory-neural-coding approach: let’s split the big black box into lots of little ones. Even so, I wish papers like this didn’t feel the need to pander so blatantly. Let the science talk!

Another scenario which encourages this type of behaviour is the academic meeting. The post-seminar mingles and afternoon coffee breaks are filled with it. Sometimes people just want to talk. They will continue batting words around for a long time past the need. Maybe it’s just the nature of the arena. Topics of note include the mind-brain problem and consciousness. Every neuroscientist is an expert. Unfortunately, most of these discussions are circular rehashes of something someone else already said a long time ago. It’s even a rare thing if I come away from a scholarly paper feeling any new light has been cast on the workings of my mind (and remember, these arguments are presumably composed following some serious contemplation). But still, it feels good to talk about cool things, right?

Of course there’s nothing harmful in all of this. People can chat about whatever they want. I do all the time. It’s just that original, useful ideas rarely emerge this way. I am believing more and more that true conceptual advances are only born following deep immersion, consideration, time and, most of all, hard work. This PhD business isn’t going to be as easy as I thought.

Refs:

• Geometric visual hallucinations, Euclidean symmetry and the functional architecture of striate cortex.
  PC Bressloff, JD Cowan, M Golubitsky, PJ Thomas, MC Wiener.
  Philos Trans R Soc Lond B Biol Sci. (2001) 356 (1407), 299-330.
  PMID: 11316482
• Reading hidden intentions in the human brain.
  JD Haynes, K Sakai, G Rees, S Gilbert, C Frith, RE Passingham.
  Curr Biol. (2007) 17(4), 323-328.
  PMID: 17291759
• Detecting and estimating signals in noisy cable structure, I: neuronal noise sources.
  A Manwani, C Koch.
  Neural Computation (1999) 11 (8), 1797-829
  PMID: 10578033
• Detecting and estimating signals in noisy cable structures, II: information theoretical analysis.
  A Manwani, C Koch.
  Neural Computation (1999) 11 (8), 1831-73
  PMID: 10578034

Monkeys mentally manoeuver mobile machines

The recent Nature paper by Velliste et al has understandably (and justifiably) received a huge amount of media coverage (e.g. this BBC report). They used electrical signals recorded from monkeys’ motor cortices to control a robotic arm. With just a few days training the animals could use it to swing a piece of food to their mouth, avoid obstacles and even adjust the movement trajectory when interrupted by the experimenter. Pretty impressive. It’s not my field, but apparently it’s the first time this technology has been demonstrated so convincingly.

The machine works by learning to associate different firing patterns in the monkey’s brain with velocities for the tip of the arm. Once the monkey learns that there is some connection between what they are thinking and what the arm is doing, all they have to do is repeat the same patterns of neural activity (thoughts) and the arm will respond accordingly. In this case, both the monkey and the computer controlling the arm are simultaneously learning how to interact.

Obviously the long-term goal for these brain-machine interfacing devices is to help humans, not monkeys. Although there is huge potential for this technology, there are a few major hurdles to overcome first – as pointed out in a comment piece by John Kalaska in the same issue of Nature.

First, to record electrical activity of any use (EEG won’t cut it), you need to implant electrodes directly into the brain. Although this is an obvious downer, you could imagine coming round to the idea if it meant being able to feed yourself again. The big problem is getting quality, stable recordings for periods longer than a few weeks. Re-operating and re-learning how to use the arm every month would be a bit of a pain – never mind the health risks. Another current problem would be having to lug around all the computer hardware (and its trained technicians) everywhere you go.

Secondly, the only current feedback the user gets from these devices is visual. You see which way the arm moves when you think of a certain motor command. Tactile information would be much more useful. You could know how hard to grip onto an apple, for instance (hard enough so it doesn’t slip, but not so hard that you crush it altogether). Of course this work is usually done by sensors in your skin, but a user could learn to interpret other signals, like visual information on a small screen, or even sound clues. Direct electrical feedback to the brain is also possible in principle, but at the moment it’s not clear how to code the information in a way that brain could understand.

Although challenging, these technical hurdles are conceivably jumpable. If so, then neuro-prosthetics will have a bright future ahead.

Refs:

• Cortical control of a prosthetic arm for self-feeding.
  Nature (2008) 453, 1098-1101
  Meel Velliste, Sagi Perel, M. Chance Spalding, Andrew S. Whitford, Andrew B. Schwartz
• Brain control of a helping hand.
  Nature (2008) 453, 994-995
  John F. Kalaska