Dendritic cytoskeleton and computation
The Dendritic Cytoskeleton as a Computational Device: An Hypothesis
Avner Priel, Jack Tuszynski & Horacion Cantiello
Dept. of Physics, University of Alberta & Harvard Medical School
In: Tuszynski, J. Ed. The Emerging Physics of Consciousness Springer ISBN-13 978-3-540-23890-4
The hypothesis in this paper is that microtubules (MTS) and actin in the dendritic cytoskeleton are active in neural computation. These proteins are suggested to interact with ion channels, microtubule associated proteins (MAPs) and kinesin. Particular importance is attached to the C-termini of the tubulin, which are suggested to exist in several conformational states, and to be reponsible for the dynamic properties of the neuroskeleton. The authors contend that ionic wave propagation along the cytoskeleton affects channel function and thence the behaviour of the dendritic tree and brain function as a whole.
Dendrites are the main site of excitatory inputs, but relatively little is still known about their functions. The activity of the particular dendritic trees, which vary greatly in shape and size, are suggested to be related to these differences. The size and complexity of dendritic trees increases with development, and this is assumed to be related to the complexity of the animal environment and memory ( 1. Kaech, Johnston, Matus ). Inputs come in at dendrite spines which are more numerous in pyramidal neurons and less so in interneurons. The number os spines and the number of excitatory inputs is clearly correlated.
The authors contends that the dynamics of the cytoskeletal structure process and deliver information to the synapse. The actin cytoskeleton is known to be related to the stability of dendritic spines ( 2. Fickova, Fischer, Landis ). Twitching of dendrite spines, which has been suggested to encode very short term memories, involves actin dynamics ( 3. Crick, Dunaevsky ). The actin part of the cytoskeleton has a key role in the formation and maintenance of synapses, and is itself remodelled by synapses. Pruning of synapses is also associated with actin ( 4. Collicos, O’ Leary, Sanes, Scott, Weimann, Zhang).
The authors argue that the extent of dendritic change in terms of growing new branches or developing new spines argues against the rather fixed quality of the traditional Hebbian model of neuronal asseblies. Recent experiments also suggest that synaptic strength is less stable than the Hebbian model suggests.
Conventionally, actin and microtubule networks have been seen as performing separate roles, with actin involved in cell movement and microtubules in transport of organelles. However more recent studies suggest that both systems have a role in what were the traditional functions of the other system ( 5. Dehmelt, Letourneau ). In fact, microtubules often grow along actin bundles. Microtubules and actin are both involved in the growth cones of cells. The authors suggest that the actin and microtubular cytoskeletons may be central to the functioning of cells.
The authors see a potentially important role for the C-termini on tubulins. It is apparent the neurons utilise MTs in some forms of cognitive processing, with both MAP2 and kinesin involved in learning and memory ( 6. Khuchua, Wong, Woolf ). It is considered likley that the transport of particular proteins and mRNA, important for synaptic development along MTs to the postsynaptic densities is important for learning. One theory put forward has been that counterions form along the length of the polymer, such as MAPs. Map2 acting as a wave guide could transfer the conformational state of a C-termini to an neighbouring MT.
Experiments show that there is a possibility of ionic wave generation along actin filaments ( 7. Cantiello, Lin ). The electrical conditions are such that it is argued that most of the ions might be tightly bound round the actin filament. This sheath of ions around the filament could mean that the it acts like an electric wire. These filaments could transmit localised waves or solitons. This actin structure has effects on the surrounding water. The water molecules reorientate themselves towards the ions and at the same time break the hydrogen bond network with neighbouring water molecules. There is then a hydration cell with water molecules orientated around an ion. An experiment has shown that actin filaments are capable of supporting ionic waves. Another experiment with actin filaments produced solitary waves very similar to those in non-linear transmission lines ( 8. Kolosick, Longren, Noguchi ). Actin filaments are abundant in dendrites and axons and this means that the experimental findings about transmissions in actin filaments has implications for signalling and ionic transport within cells. There is extensive new information showing that actin filaments are linked to ion channels ( 9. Chasan, Janmey ). Actin filaments can change their configuration and it is speculated that ionic waves may be involved in this process. In neurons actin is mainly concentrated in the synaptic areas, and it is considered feasible that electrical signals through actin may help to trigger neurotransmitter release, and that in the dendrites it may be involved in the pos-synaptic response. Kaech et al ( 1. ) showed that anesthetics inhibited the actin response in dendrites. The authors expect ionic waves along actin filaments to be shown to have a broad range of effects. They say that the core of their theory is the propagation of ionic waves along actin filaments, MAPs that interact with them and C-termini on tubulins. The interaction between these and membrane components such as ion channels could produce previously undetected modulatory effects on synaptic connections.
Microtubules and actin filaments are interconnected, and actin filaments are connected to ion channels. Actin bundles bind to post-synaptic densities in dendrites and dendrite spines. At the other end the actin binds to microtubules. Actin also binds to ion channels. It is envisaged that the electrical reponsiveness of the neuron may be regulated via this cytoskeletal connection to the ion channels.
In this model, microtubules in dendrites receive signals from synapses via ion waves propagated along actin filaments that are connected to microtubules by MAP2. The inputs influence the evolution of an existing system. The microtubules develop the inputs by means of the changing conformation of the C-termini, with some operations recurrent where MAPs connected to adjacent MTs.
Finally, the MTs produce a read out to ion channels often via wave propagation along actin filaments, and are suggested to regulate voltage sensitive ion channels. This in turn regulates the axon hillock and the output of axons potentials by changing the distribution of open and closed ion channels.
The information processing in dendrites is assisted by their special lay out with short microtubules of mixed polarity connected by MAP2. It is considered possible that there could be a Hebbian-type system in which frequent activity in parts of the microtubule could produce a higher or lower actin filament density, which would constitute memory/learning. Johnson and Byerly ( 9. ) showed that agents that modified the cytoskeleton also alter calcium ion activity in some neurons. Potassium channels have been shown to be controlled by disuption of actin filaments ( 10. Maguire ). At the close of the chapter, the authors stress their key finding, which is that MTs, MAPs and actin filaments support ionic waves, and their hypothesis that these ionic waves may have a role in neural function.
1) Kaech, S., et al (1999) Proceedings of the National Academy of Sciences, USA, 96, 10433-10437 + Kaech, S. et al (2001) Proceedings of the National Academy of Sciences, USA, 98, pp. 7086-7092 + Johnston, D. et al (1996) Annual Review of Neuroscience, 19, pp. 165-86 + Johnston, D. et al (1992) Annual Review of Physiology, 54, pp. 489-505 + Matus, A. (2000) Science, 290, pp.754-758 + Matus, A. et al (1982) Proceedings of the National Academy of Science, USA, 79, pp. 7590-7594
2) Fifkova, E. & Delay, R. (1982) Journal of Cell Biology, 95, pp. 345-50 + Fischer, M. et al (1998) Neuron, 20, pp. 847-54 + Fischer, M. et al (2000) Nature Neuroscience, 3, pp. 887-894 + Landis, D. & Reese, T. (1983) Journal of Cell Biology, 97, pp. 1169-1178
3) Crick, F. (1982) Trends in Neuroscience, 5, pp. 44-46 + Dunaevsky, A. et al (1999) Proceedings of the National Academy of Sciences, USA, 96, pp. 13438-13443
4) Colicos, M. et al (2001) Cell, 30, pp. 605-616 + O’ Leary, D. & Koester, S. (1993) Neuron, 10, pp. 991 + Sanes, J. & Lichtman, J. (1999) Annual Review of Neuroscience, 22 + Scott, E. & Luo, L. (2001) National Neuroscience, 4, pp. 359-365 + Wang, Y. (1985) Journal of Cell Biology, 78, pp. 1955-1964 + Zhang, W. & Benson, D. (2001) Journal of Neuroscience, 15, pp. 5169-5181 + Weimann, J. et al (1999) Neuron, 24, pp. 819-31
5) Dehmelt, L. & Halpain, S. (2004) Journal of Neurobiology, 58, pp. 18-33 + Dehmelt, L. et al (2003) Journal of Neuroscience, 23, pp. 9479-90 + Letorneau, P. (1996) Perspect. Dev. Neurobiology, 4, pp. 111-123 + Letorneau, P. & Ressler, A. (1984) Journal of Cell Biology, 98, pp. 1355-1362
6) Khuchua, Z. et al (2003) Neuroscience, 119 pp. 101-6 + Woolf, Nancy et al (1999) Brain Research, 821, pp. 241-9 + Wong, R. et al (2002) Proceedings of the National Academy of Sciences, USA, 99, pp. 14500-14505
7) Cantiello, H. et al (1991) Biophysics Journal, 59, pp. 1284-9 + Lin, E. & Cantiello, H. (1993) Biophysics Journal, 65, pp. 1371-8
8) Kolosick, J. et al (1974) Proceedings IEEE, 62, pp. 578-581 + Lonngren, K. (1978) In: Lonngren, K. & Scott, A. Eds Solitons in Action Academic Press + Lonngren, K. et al (1975) IEEE Trans Circuits and Systems, CAS-22, pp. 376-78 + Noguchi, A. (1974) Elec. and Comm. In Japan, 57-A, pp. 9-13
9) Chasan, B. et al (2002) Eur. Biophysical Journal, 30, pp. 617-624 + Janmey, P. (1998) Physiology Review, 78, pp. 763-81